#+TITLE: org-babel --- facilitating communication between programming languages and people #+SEQ_TODO: TODO PROPOSED STARTED | DONE DEFERRED REJECTED #+OPTIONS: H:3 num:nil toc:t #+STARTUP: oddeven hideblocks Through Org-Babel Org-Mode can communicate with programming languages. Code contained in source-code blocks can be evaluated and data can pass seamlessly between different programming languages, Org-Mode constructs (tables, file links, example text) and interactive comint buffers. In this document: - The [[* Introduction][Introduction]] :: provides a brief overview of the design and use of Org-Babel including tutorials and examples. - In [[* Getting started][Getting Started]] :: find instructions for installing org-babel into your emacs configuration. - The [[* Tasks][Tasks]] :: section contains current and past tasks roughly ordered by TODO state, then importance or date-completed. This would be a good place to suggest ideas for development. - The [[* Bugs][Bugs]] :: section contains bug reports. - The [[* Tests][Tests]] :: section consists of a large table which can be evaluated to run Org-Babel's functional test suite. This provides a good overview of the current functionality with pointers to example source blocks. - The [[* Sandbox][Sandbox]] :: demonstrates much of the early/basic functionality through commented source-code blocks. Also see the [[file:library-of-babel.org][Library of Babel]] an extensible collection of ready-made and easily-shortcut-callable source-code blocks for handling common tasks. * Introduction Org-Babel enables *communication* between programming languages and between people. Org-Babel provides: - communication between programs :: Data passes seamlessly between different programming languages, Org-Mode constructs (tables, file links, example text) and interactive comint buffers. - communication between people :: Data and calculations are embedded in the same document as notes explanations and reports. ** communication between programs Org-Mode supports embedded blocks of source code (in any language) inside of Org documents. Org-Babel allows these blocks of code to be executed from within Org-Mode with natural handling of their inputs and outputs. *** simple execution with both scalar, file, and table output *** reading information from tables *** reading information from other source blocks (disk usage in your home directory) This will work for Linux and Mac users, not so sure about shell commands for windows users. To run place the cursor on the =#+begin_src= line of the source block labeled directory-pie and press =\C-c\C-c=. #+srcname: directories #+begin_src bash :results replace cd ~ && du -sc * |grep -v total #+end_src #+resname: directories | 64 | "Desktop" | | 11882808 | "Documents" | | 8210024 | "Downloads" | | 879800 | "Library" | | 57344 | "Movies" | | 7590248 | "Music" | | 5307664 | "Pictures" | | 0 | "Public" | | 152 | "Sites" | | 8 | "System" | | 56 | "bin" | | 3274848 | "mail" | | 5282032 | "src" | | 1264 | "tools" | #+srcname: directory-pie #+begin_src R :var dirs = directories pie(dirs[,1], labels = dirs[,2]) #+end_src *** operations in/on tables #+tblname: grades-table | student | grade | letter | |---------+-------+--------| | 1 | 99 | A | | 2 | 59 | F | | 3 | 75 | C | | 4 | 15 | F | | 5 | 7 | F | | 6 | 13 | F | #+TBLFM: $2='(sbe random-score-generator)::$3='(sbe assign-grade (score $2)) #+srcname: assign-grade #+begin_src ruby :var score=99 case score when 0..59: "F" when 60..69: "D" when 70..79: "C" when 80..89: "B" when 90..100: "A" else "Invalid Score" end #+end_src #+srcname: random-score-generator #+begin_src ruby rand(100) #+end_src #+srcname: show-distribution #+begin_src R :var grades=grades-table hist(grades[,2]) #+end_src ** communication between people Quick overview of Org-Mode's exportation abilities, with links to the online Org-Mode documentation, a focus on source-code blocks, and the exportation options provided by Org-Babel. *** Interactive tutorial This would demonstrate applicability to Reproducible Research, and Literate Programming. *** Tests embedded in documentation org-babels own functional tests are contained in a large org-mode table, allowing the test suite to be run be evaluation of the table and the results to be collected in the same table. *** Emacs initialization files stored in Org-Mode buffers Once org-babel-tangle is completed this could be a very compelling use case. ** features *** code evaluation (comint buffer sessions and external processes) There are two main ways to evaluate source blocks with org-babel. - external :: By default (if the =:session= header argument is not present) all source code blocks are evaluated in external processes. In these cases an external process is used to evaluate the source-code blocks. - session :: Session based evaluation uses persistent sessions in comint buffers. Sessions can be used across multiple source blocks setting and accessing variables in the global environment. Evaluating source blocks in sessions also allows for interaction with the code. To jump to the session of a source block use the `org-babel-pop-to-session' command or press =M-[down]= while inside of a source code block. When called with a prefix argument `org-babel-pop-to-session' will evaluate all header arguments before jumping to the source-code block. *** results (values and outputs) Either the *value* or the *output* of source code blocks can be collected after evaluation. - value :: The default way to collect results from a source-code block is to return the value of the last statement in the block. This can be thought of as the return value of the block. In this case any printed output of the block is ignored. This can be though of a similar to a "functional" value of evaluation. - output :: Another way of generating results from a source-code block is to collect the output generated by the execution of the block. In this case all printed output is collected throughout the execution of the block. This can be thought of as similar to a "script" style of evaluation. * Getting started Add the following lines to your .emacs, replacing the path as appropriate. A good place to check that things are up and running would then be [[#sandbox][the sandbox]]. #+begin_src emacs-lisp (add-to-list 'load-path "/path/to/org-babel/lisp") (require 'org-babel-init) #+end_src * Tasks [27/42] ** TODO support for working with =*Org Edit Src Example*= buffers [2/4] *** TODO optionally evaluate header references when we switch to =*Org Edit Src*= buffer That seems to imply that the header references need to be evaluated and transformed into the target language object when we hit C-c ' to enter the *Org Edit Src* buffer [DED] Good point, I heartily agree that this should be supported [Eric] (or at least before the first time we attempt to evaluate code in that buffer -- I suppose there might be an argument for lazy evaluation, in case someone hits C-c ' but is "just looking" and not actually evaluating anything.) Of course if evaluating the reference is computationally intensive then the user might have to wait before they get the *Org Edit Src* buffer. [DED] I fear that it may be hard to anticipate when the references will be needed, some major-modes do on-the-fly evaluation while the buffer is being edited. I think that we should either do this before the buffer is opened or not at all, specifically I think we should resolve references if the user calls C-c ' with a prefix argument. Does that sound reasonable? [Eric] Yes [Dan] [Dan] So now that we have org-src-mode and org-src-mode-hook, I guess org-babel should do this by using the hook to make sure that, when C-c C-' is issued on a source block, any references are resolved and assignments are made in the appropriate session. *** TODO set buffer-local-process variables appropriately [DED] I think something like this would be great. You've probably already thought of this, but just to note it down: it would be really nice if org-babel's notion of a buffer's 'session/process' played nicely with ESS's notion of the buffer's session/process. ESS keeps the current process name for a buffer in a buffer-local variable ess-local-process-name. So one thing we will probably want to do is make sure that the *Org Edit Src Example* buffer sets that variable appropriately. [DED] I had not thought of that, but I agree whole heartedly. [Eric] Once this is done every variable should be able to dump regions into their inferior-process buffer using major-mode functions. *** DEFERRED send code to inferior process Another thought on this topic: I think we will want users to send chunks of code to the interpreter from within the *Org Edit Src* buffer, and I think that's what you have in mind already. In ESS that is done using the ess-eval-* functions. [DED] I think we can leave this up to the major-mode in the source code buffer, as almost every source-code major mode will have functions for doing things like sending regions to the inferior process. If anything we might need to set the value of the buffer local inferior process variable. [Eric] *** DONE some possible requests/proposed changes for Carsten [4/4] While I remember, some possible requests/proposed changes for Carsten come to mind in that regard: **** DONE Remap C-x C-s to save the source to the org buffer? I've done this personally and I find it essential. I'm using #+begin_src emacs-lisp (defun org-edit-src-save () "Update the parent org buffer with the edited source code, save the parent org-buffer, and return to the source code edit buffer." (interactive) (let ((p (point))) (org-edit-src-exit) (save-buffer) (org-edit-src-code) (goto-char p))) (define-key org-exit-edit-mode-map "\C-x\C-s" 'org-edit-src-save) #+end_src which seems to work. I think this is great, but I think it should be implemented in the org-mode core **** DEFERRED Rename buffer and minor mode? Something shorter than *Org Edit Src Example* for the buffer name. org-babel is bringing org's source code interaction to a level of maturity where the 'example' is no longer appropriate. And if further keybindings are going to be added to the minor mode then maybe org-edit-src-mode is a better name than org-exit-edit-mode. Maybe we should name the buffer with a combination of the source code and the session. I think that makes sense. [ES] Are you also suggesting a new org-edit-src minor mode? [DED] org-exit-edit-mode is a minor mode that already exists: Minor mode installing a single key binding, "C-c '" to exit special edit. org-edit-src-save now has a binding in that mode, so I guess all I'm saying at this stage is that it's a bit of a misnomer. But perhaps we will also have more functionality to add to that minor mode, making it even more of a misnomer. Perhaps something like org-src-mode would be better. **** DONE Changed minor mode name and added hooks **** DEFERRED a hook called when the src edit buffer is created This should be implemented in the org-mode core ** TODO resolve references to other org buffers/files This would allow source blocks to call upon tables, source-blocks, and results in other org buffers/files. See... - [[file:lisp/org-babel-ref.el::TODO%20allow%20searching%20for%20names%20in%20other%20buffers][org-babel-ref.el:searching-in-other-buffers]] - [[file:lisp/org-babel.el::defun%20org-babel%20find%20named%20result%20name][org-babel.el#org-babel-find-named-result]] ** TODO resolve references to other non-org files - tabular data in .csv, .tsv etc format - files of interpreted code: anything stopping us giving such files similar status to a source code block? - Would be nice to allow org and non-org files to be remote ** TODO figure out how to handle errors during evaluation R has a try function, with error handling, along the lines of python. I bet ruby does too. Maybe more of an issue for functional style; in my proposed scripting style the error just gets dumped to the org buffer and the user is thus alerted. ** TODO figure out how to handle graphic output This is listed under [[* graphical output][graphical output]] in out objectives. This should take advantage of the =:results file= option, and languages which almost always produce graphical output should set =:results file= to true by default. That would handle placing these results in the buffer. Then if there is a combination of =silent= and =file= =:results= headers we could drop the results to a temp buffer and pop open that buffer... ** TODO Finalise behaviour regarding vector/scalar output *** DONE Stop spaces causing vector output This simple example of multilingual chaining produces vector output if there are spaces in the message and scalar otherwise. [Not any more] #+begin_src R :var msg=msg-from-python paste(msg, "und R", sep=" ") #+end_src #+resname: : org-babel speaks elisp y python und R #+srcname: msg-from-python #+begin_src python :var msg=msg-from-elisp msg + " y python" #+end_src #+srcname: msg-from-elisp #+begin_src emacs-lisp :var msg="org-babel speaks" (concat msg " elisp") #+end_src ** STARTED share org-babel [1/4] how should we share org-babel? *** DONE post to org-mode *** TODO post to ess mailing list *** TODO create a org-babel page on worg *** TODO create a short screencast demonstrating org-babel in action *** examples we need to think up some good examples **** interactive tutorials This could be a place to use [[* org-babel assertions][org-babel assertions]]. for example the first step of a tutorial could assert that the version of the software-package (or whatever) is equal to some value, then source-code blocks could be used with confidence (and executed directly from) the rest of the tutorial. **** answering a text-book question w/code example org-babel is an ideal environment enabling both the development and demonstrationg of the code snippets required as answers to many text-book questions. **** something using tables maybe something along the lines of calculations from collected grades **** file sizes Maybe something like the following which outputs sizes of directories under the home directory, and then instead of the trivial =emacs-lisp= block we could use an R block to create a nice pie chart of the results. #+srcname: sizes #+begin_src bash :results replace du -sc ~/* #+end_src #+begin_src emacs-lisp :var sizes=sizes :results replace (mapcar #'car sizes) #+end_src ** TODO command line execution Allow source code blocks to be called form the command line. This will be easy using the =sbe= function in [[file:lisp/org-babel-table.el][org-babel-table.el]]. This will rely upon [[* resolve references to other buffers][resolve references to other buffers]]. ** TODO inline source code blocks [3/5] Like the =\R{ code }= blocks not sure what the format should be, maybe just something simple like =src_lang[]{}= where lang is the name of the source code language to be evaluated, =[]= is optional and contains any header arguments and ={}= contains the code. (see [[* (sandbox) inline source blocks][the-sandbox]]) *** DONE evaluation with \C-c\C-c Putting aside the header argument issue for now we can just run these with the following default header arguments - =:results= :: silent - =:exports= :: results *** DONE inline exportation Need to add an interblock hook (or some such) through org-exp-blocks *** DONE header arguments We should make it possible to use header arguments. *** TODO fontification we should color these blocks differently *** TODO refine html exportation should use a span class, and should show original source in tool-tip ** TODO formulate general rules for handling vectors and tables / matrices with names This is non-trivial, but may be worth doing, in particular to develop a nice framework for sending data to/from R. *** Notes In R, indexing vector elements, and rows and columns, using strings rather than integers is an important part of the language. - elements of a vector may have names - matrices and data.frames may have "column names" and "row names" which can be used for indexing - In a data frame, row names *must* be unique Examples #+begin_example > # a named vector > vec <- c(a=1, b=2) > vec["b"] b 2 > mat <- matrix(1:4, nrow=2, ncol=2, dimnames=list(c("r1","r2"), c("c1","c2"))) > mat c1 c2 r1 1 3 r2 2 4 > # The names are separate from the data: they do not interfere with operations on the data > mat * 3 c1 c2 r1 3 9 r2 6 12 > mat["r1","c2"] [1] 3 > df <- data.frame(var1=1:26, var2=26:1, row.names=letters) > df$var2 [1] 26 25 24 23 22 21 20 19 18 17 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 > df["g",] var1 var2 g 7 20 #+end_example So it's tempting to try to provide support for this in org-babel. For example - allow R to refer to columns of a :var reference by their names - When appropriate, results from R appear in the org buffer with "named columns (and rows)" However none (?) of the other languages we are currently supporting really have a native matrix type, let alone "column names" or "row names". Names are used in e.g. python and perl to refer to entries in dicts / hashes. It currently seems to me that support for this in org-babel would require setting rules about when org tables are considered to have named columns/fields, and ensuring that (a) languages with a notion of named columns/fields use them appropriately and (b) languages with no such notion do not treat then as data. - Org allows something that *looks* like column names to be separated by a hline - Org also allows a row to *function* as column names when special markers are placed in the first column. An hline is unnecessary (indeed hlines are purely cosmetic in org [correct?] - Org does not have a notion of "row names" [correct?] The full org table functionality exeplified [[http://orgmode.org/manual/Advanced-features.html#Advanced-features][here]] has features that we would not support in e.g. R (like names for the row below). *** Initial statement: allow tables with hline to be passed as args into R This doesn't seem to work at the moment (example below). It would also be nice to have a natural way for the column names of the org table to become the column names of the R data frame, and to have the option to specify that the first column is to be used as row names in R (these must be unique). But this might require a bit of thinking about. #+TBLNAME: egtable | col1 | col2 | col3 | |------+---------+------| | 1 | 2 | 3 | | 4 | schulte | 6 | #+TBLNAME: egtable2 | 1 | 2 | 3 | | 4 | schulte | 6 | #+begin_src R var tabel=egtable tabel #+end_src #+resname: | "col1" | "col2" | "col3" | |--------+-----------+--------| | 1 | 2 | 3 | | 4 | "schulte" | 6 | Another example is in the [[*operations%20in%20on%20tables][grades example]]. ** TODO re-implement helper functions from org-R *** Initial statement [Eric] Much of the power of org-R seems to be in it's helper functions for the quick graphing of tables. Should we try to re-implement these functions on top of org-babel? I'm thinking this may be useful both to add features to org-babel-R and also to potentially suggest extensions of the framework. For example one that comes to mind is the ability to treat a source-code block like a function which accepts arguments and returns results. Actually this can be it's own TODO (see [[* source blocks as functions][source blocks as functions]]). *** Objectives [Dan] - We want to provide convenient off-the-shelf actions (e.g. plotting data) that make use of our new code evaluation environment but do not require any actual coding. *** Initial Design proposal [Dan] - *Input data* will be specified using the same mechanism as :var references, thus the input data may come from a table, or another source block, and it is initially available as an elisp data structure. - We introduce a new #+ line, e.g. #+BABELDO. C-c C-c on that line will apply an *action* to the referenced data. - *Actions correspond to source blocks*: our library of available actions will be a library of org-babel source blocks. Thus the code for executing an action, and the code for dealing with the output of the action will be the same code as for executing source blocks in general - Optionally, the user can have the relevant source block inserted into the org buffer after the (say) #+BABELDO line. This will allow the user to fine tune the action by modifying the code (especially useful for plots). - So maybe a #+BABELDO line will have header args - :data (a reference to a table or source code block) - :action (or should that be :srcname?) which will be something like :action pie-chart, referring to a source block which will be executed with the :data referent passed in using a :var arg. - :showcode or something controlling whether to show the code *** Modification to design I'm implementing this, at least initially, as a new interpreter named 'babel', which has an empty body. 'babel' blocks take a :srcname header arg, and look for the source-code block with that name. They then execute the referenced block, after first appending their own header args on to the target block's header args. If the target block is in the library of babel (a.o.t. e.g. the current buffer), then the code in the block will refer to the input data with a name dictated by convention (e.g. __data__ (something which is syntactically legal in all languages...). Thus the babel block will use a :var __data__ = whatever header arg to reference the data to be plotted. *** Current design This is covered by the [[file:library-of-babel.org][Library of Babel]], which will contain ready-made source blocks designed to carry out useful common tasks. ** PROPOSED Are we happy with current behaviour regarding vector/scalar output? This simple example of multilingual chaining produces vector output if there are spaces in the message and scalar otherwise. #+begin_src R :var msg=msg-from-python paste(msg, "und_R", sep="_") #+end_src #+srcname: msg-from-python #+begin_src python :var msg=msg-from-elisp msg + "_y_python" #+end_src #+srcname: msg-from-elisp #+begin_src emacs-lisp :var msg="org-babel_speaks" (concat msg "_elisp") #+end_src ** PROPOSED conversion between org-babel and noweb (e.g. .Rnw) format I haven't thought about this properly. Just noting it down. What Sweave uses is called "R noweb" (.Rnw). I found a good description of noweb in the following article (see the [[http://www.cs.tufts.edu/~nr/pubs/lpsimp.pdf][pdf]]). I think there are two parts to noweb, the construction of documentation and the extraction of source-code (with notangle). *documentation*: org-mode handles all of our documentation needs in a manner that I believe is superior to noweb. *source extraction* At this point I don't see anyone writing large applications with 100% of the source code contained in org-babel files, rather I see org-babel files containing things like - notes with active code chunks - interactive tutorials - requirements documents with code running test suites - and of course experimental reports with the code to run the experiment, and perform analysis Basically I think the scope of the programs written in org-babel (at least initially) will be small enough that it wont require the addition of a tangle type program to extract all of the source code into a running application. On the other hand, since we already have named blocks of source code which reference other blocks on which they rely, this shouldn't be too hard to implement either on our own, or possibly relying on something like noweb/notangle. ** PROPOSED support for passing paths to files between source blocks Maybe this should be it's own result type (in addition to scalars and vectors). The reason being that some source-code blocks (for example ditaa or anything that results in the creation of a file) may want to pass a file path back to org-mode which could then be inserted into the org-mode buffer as a link to the file... This would allow for display of images upon export providing functionality similar to =org-exp-blocks= only in a more general manner. ** DEFERRED use textConnection to pass tsv to R? When passing args from the org buffer to R, the following route is used: arg in buffer -> elisp -> tsv on file -> data frame in R. I think it would be possible to avoid having to write to file by constructing an R expression in org-babel-R-assign-elisp, something like this #+begin_src emacs-lisp (org-babel-R-input-command (format "%s <- read.table(textConnection(\"%s\"), sep=\"\\t\", as.is=TRUE)" name (orgtbl-to-tsv value '(:sep "\t" :fmt org-babel-R-quote-tsv-field)))) #+end_src I haven't tried to implement this yet as it's basically just fiddling with something that works. The only reason for it I can think of would be efficiency and I haven't tested that. This Didn't work after an initial test. I still think this is a good idea (I also think we should try to do something similar when writing out results frmo R to elisp) however as it wouldn't result in any functional changes I'm bumping it down to deferred for now. [Eric] for quick tests #+tblname: quick-test | 1 | 2 | 3 | #+srcname: quick-test-src-blk #+begin_src R :var vec=quick-test mean(mean(vec)) #+end_src : 2 ** DEFERRED re-implement R evaluation using ess-command or ess-execute I don't have any complaints with the current R evaluation code or behaviour, but I think it would be good to use the ESS functions from a political point of view. Plus of course it has the normal benefits of an API (insulates us from any underlying changes etc). [DED] I'll look into this. I believe that I looked at and rejected these functions initially but now I can't remember why. I agree with your overall point about using API's where available. I will take a look back at these and either switch to using the ess commands, or at least articulate under this TODO the reasons for using our custom R-interaction commands. [Eric] ess-execute Lets just replace =org-babel-R-input-command= with =ess-execute=. I tried this, and although it works in some situations, I find that =ess-command= will often just hang indefinitely without returning results. Also =ess-execute= will occasionally hang, and pops up the buffer containing the results of the command's execution, which is undesirable. For now these functions can not be used. Maybe someone more familiar with the ESS code can recommend proper usage of =ess-command= or some other lower-level function which could be used in place of [[file:lisp/org-babel-R.el::defun%20org-babel%20R%20input%20command%20command][org-babel-R-input-command]]. *** ess functions #+begin_quote ess-command (ess-command COM &optional BUF SLEEP NO-PROMPT-CHECK) Send the ESS process command COM and delete the output from the ESS process buffer. If an optional second argument BUF exists save the output in that buffer. BUF is erased before use. COM should have a terminating newline. Guarantees that the value of .Last.value will be preserved. When optional third arg SLEEP is non-nil, `(sleep-for (* a SLEEP))' will be used in a few places where `a' is proportional to `ess-cmd-delay'. #+end_quote #+begin_quote ess-execute (ess-execute COMMAND &optional INVERT BUFF MESSAGE) Send a command to the ESS process. A newline is automatically added to COMMAND. Prefix arg (or second arg INVERT) means invert the meaning of `ess-execute-in-process-buffer'. If INVERT is 'buffer, output is forced to go to the process buffer. If the output is going to a buffer, name it *BUFF*. This buffer is erased before use. Optional fourth arg MESSAGE is text to print at the top of the buffer (defaults to the command if BUFF is not given.) #+end_quote *** out current setup 1) The body of the R source code block is wrapped in a function 2) The function is called inside of a =write.table= function call writing the results to a table 3) The table is read using =org-table-import= ** DEFERRED Rework Interaction with Running Processes [2/5] *** DONE robust to errors interrupting execution #+srcname: long-runner-ruby #+begin_src ruby :results silent sleep(10) :patton_is_an_grumpy #+end_src *** DEFERRED use =C-g= keyboard-quit to push processing into the background This may be possible using the `run-with-timer' command. I have no idea how this could work... #+srcname: long-runner-ruby #+begin_src ruby :results silent sleep(10) :patton_is_an_grumpy #+end_src *** TODO ability to select which of multiple sessions is being used Increasingly it is looking like we're going to want to run all source code blocks in comint buffer (sessions). Which will have the benefits of 1) allowing background execution 2) maintaining state between source-blocks - allowing inline blocks w/o header arguments **** R sessions (like ess-switch-process in .R buffers) Maybe this could be packaged into a header argument, something like =:R_session= which could accept either the name of the session to use, or the string =prompt=, in which case we could use the =ess-switch-process= command to select a new process. *** TODO evaluation of shell code as background process? After C-c C-c on an R code block, the process may appear to block, but C-g can be used to reclaim control of the .org buffer, without interrupting the R evalution. However I believe this is not true of bash/sh evaluation. [Haven't tried other languages] Perhaps a solution is just to background the individual shell commands. The other languages (aside from emacs lisp) are run through the shell, so if we find a shell solution it should work for them as well. Adding an ampersand seems to be a supported way to run commands in the background (see [[http://www.emacswiki.org/emacs/ExecuteExternalCommand#toc4][external-commands]]). Although a more extensible solution may involve the use of the [[elisp:(progn (describe-function 'call-process-region) nil)][call-process-region]] function. Going to try this out in a new file [[file:lisp/org-babel-proc.el][org-babel-proc.el]]. This should contain functions for asynchronously running generic shell commands in the background, and then returning their input. **** partial update of org-mode buffer The sleekest solution to this may be using a comint buffer, and then defining a filter function which would incrementally interpret the results as they are returned, including insertion into the org-mode buffer. This may actually cause more problems than it is worth, what with the complexities of identifying the types of incrementally returned results, and the need for maintenance of a process marker in the org buffer. **** 'working' spinner It may be nice and not too difficult to place a spinner on/near the evaluating source code block *** TODO conversion of output from interactive shell, R (and python) sessions to org-babel buffers [DED] This would be a nice feature I think. Although an org-babel purist would say that it's working the wrong way round... After some interactive work in a *R* buffer, you save the buffer, maybe edit out some lines, and then convert it to org-babel format for posterity. Same for a shell session either in a *shell* buffer, or pasted from another terminal emulator. And python of course. ** DEFERRED improve the source-block snippet any real improvement seems somewhat beyond the ability of yasnippet for now. [[file:~/src/emacs-starter-kit/src/snippets/text-mode/rst-mode/chap::name%20Chapter%20title][file:~/src/emacs-starter-kit/src/snippets/text-mode/rst-mode/chap::name Chapter title]] #+begin_example ,#name : Chapter title ,# -- ${1:Chapter} ${1:$(make-string (string-width text) ?\=)} $0 #+end_example [[file:snippets/org-mode/sb][sb -- snippet]] waiting for guidance from those more familiar with yasnippets ** DONE add a function to jump to a source-block by name I've had an initial stab at that in org-babel-find-named-block (library-of-babel branch). At the same time I introduced org-babel-named-src-block-regexp, to match src-blocks with srcname. This is now working with the command `org-babel-goto-named-source-block', all we need is a good key binding. ** DONE add =:none= session argument (for purely functional execution) [4/4] This would allow source blocks to be run in their own new process - These blocks could then also be run in the background (since we can detach and just wait for the process to signal that it has terminated) - We wouldn't be drowning in session buffers after running the tests - we can re-use much of the session code to run in a more /functional/ mode While session provide a lot of cool features, like persistent environments, [[* DONE function to bring up inferior-process buffer][pop-to-session]], and hints at exportation for org-babel-tangle, they also have some down sides and I'm thinking that session-based execution maybe shouldn't be the default behavior. Down-sides to sessions - *much* more complicated than functional evaluation - maintaining the state of the session has weird issues - waiting for evaluation to finish - prompt issues like [[* TODO weird escaped characters in shell prompt break shell evaluation][shell-prompt-escapes-bug]] - can't run in background - litter emacs with session buffers *** DONE ruby #+srcname: ruby-task-no-session #+begin_src ruby :results replace output puts :eric puts :schulte [1, 2, 3] #+end_src #+resname: ruby-task-no-session | "eric" | | "schulte" | *** DONE python #+srcname: task-python-none-session #+begin_src python :session none :results replace value print 'something' print 'output' [1, 2, 3] #+end_src #+resname: task-python-none-session | 1 | 2 | 3 | *** DONE sh #+srcname: task-session-none-sh #+begin_src sh :results replace echo "first" echo "second" #+end_src #+resname: task-session-none-sh | "first" | | "second" | *** DONE R #+srcname: task-no-session-R #+begin_src R :results replace output a <- 8 b <- 9 a + b b - a #+end_src #+resname: task-no-session-R | "[1]" | 17 | | "[1]" | 1 | ** DONE fully purge org-babel-R of direct comint interaction try to remove all code under the [[file:lisp/org-babel-R.el::functions%20for%20evaluation%20of%20R%20code][;; functions for evaluation of R code]] line ** DONE Create objects in top level (global) environment [5/5] *sessions* *** initial requirement statement [DED] At the moment, objects created by computations performed in the code block are evaluated in the scope of the code-block-function-body and therefore disappear when the code block is evaluated {unless you employ some extra trickery like assign('name', object, env=globalenv()) }. I think it will be desirable to also allow for a style wherein objects that are created in one code block persist in the R global environment and can be re-used in a separate block. This is what Sweave does, and while I'm not saying we have to be the same as Sweave, it wouldn't be hard for us to provide the same behaviour in this case; if we don't, we risk undeservedly being written off as an oddity by some. IOW one aspect of org-babel is that of a sort of functional meta-programming language. This is crazy, in a very good way. Nevertheless, wrt R I think there's going to be a lot of value in providing for a working style in which the objects are stored in the R session, rather than elisp/org buffer. This will be a very familiar working style to lots of people. There are no doubt a number of different ways of accomplishing this, the simplest being a hack like adding #+begin_src R for(objname in ls()) assign(objname, get(objname), envir=globalenv()) #+end_src to the source code block function body. (Maybe wrap it in an on.exit() call). However this may deserve to be thought about more carefully, perhaps with a view to having a uniform approach across languages. E.g. shell code blocks have the same semantics at the moment (no persistence of variables across code blocks), because the body is evaluated in a new bash shell process rather than a running shell. And I guess the same is true for python. However, in both these cases, you could imagine implementing the alternative in which the body is evaluated in a persistent interactive session. It's just that it's particularly natural for R, seeing as both ESS and org-babel evaluate commands in a single persistent R session. *** sessions [Eric] Thanks for bringing this up. I think you are absolutely correct that we should provide support for a persistent environment (maybe called a *session*) in which to evaluate code blocks. I think the current setup demonstrates my personal bias for a functional style of programming which is certainly not ideal in all contexts. While the R function you mention does look like an elegant solution, I think we should choose an implementation that would be the same across all source code types. Specifically I think we should allow the user to specify an optional *session* as a header variable (when not present we assume a default session for each language). The session name could be used to name a comint buffer (like the *R* buffer) in which all evaluation would take place (within which variables would retain their values --at least once I remove some of the functional method wrappings currently in place-- ). This would allow multiple environments to be used in the same buffer, and once this setup was implemented we should be able to fairly easily implement commands for jumping between source code blocks and the related session buffers, as well as for dumping the last N commands from a session into a new or existing source code block. Please let me know if you foresee any problems with this proposed setup, or if you think any parts might be confusing for people coming from Sweave. I'll hopefully find some time to work on this later in the week. *** can functional and interpreted/interactive models coexist? Even though both of these use the same =*R*= buffer the value of =a= is not preserved because it is assigned inside of a functional wrapper. #+srcname: task-R-sessions #+begin_src R a <- 9 b <- 21 a + b #+end_src #+srcname: task-R-same-session #+begin_src R a #+end_src This functional wrapper was implemented in order to efficiently return the results of the execution of the entire source code block. However it inhibits the evaluation of source code blocks in the top level, which would allow for persistence of variable assignment across evaluations. How can we allow *both* evaluation in the top level, and efficient capture of the return value of an entire source code block in a language independent manner? Possible solutions... 1) we can't so we will have to implement two types of evaluation depending on which is appropriate (functional or imperative) 2) we remove the functional wrapper and parse the source code block into it's top level statements (most often but not always on line breaks) so that we can isolate the final segment which is our return value. 3) we add some sort of "#+return" line to the code block 4) we take advantage of each languages support for meta-programming through =eval= type functions, and use said to evaluate the entire blocks in such a way that their environment can be combined with the global environment, and their results are still captured. 5) I believe that most modern languages which support interactive sessions have support for a =last_result= type function, which returns the result of the last input without re-calculation. If widely enough present this would be the ideal solution to a combination of functional and imperative styles. None of these solutions seem very desirable, but for now I don't see what else would be possible. Of these options I was leaning towards (1) and (4) but now believe that if it is possible option (5) will be ideal. **** (1) both functional and imperative evaluation Pros - can take advantage of built in functions for sending regions to the inferior process - retains the proven tested and working functional wrappers Cons - introduces the complication of keeping track of which type of evaluation is best suited to a particular context - the current functional wrappers may require some changes in order to include the existing global context **** (4) exploit language meta-programming constructs to explicitly evaluate code Pros - only one type of evaluation Cons - some languages may not have sufficient meta-programming constructs **** (5) exploit some =last_value= functionality if present Need to ensure that most languages have such a function, those without will simply have to implement their own similar solution... | language | =last_value= function | |------------+-----------------------------| | R | .Last.value | | ruby | _ | | python | _ | | shell | see [[* last command for shells][last command for shells]] | | emacs-lisp | see [[* emacs-lisp will be a special case][special-case]] | #+srcname: task-last-value #+begin_src ruby 82 + 18 #+end_src ***** last command for shells Do this using the =tee= shell command, and continually pipe the output to a file. Got this idea from the following [[http://linux.derkeiler.com/Mailing-Lists/Fedora/2004-01/0898.html][email-thread]]. suggested from mailing list #+srcname: bash-save-last-output-to-file #+begin_src sh while read line do bash -c "$line" | tee /tmp/last.out1 mv /tmp/last.out1 /tmp/last.out done #+end_src another proposed solution from the above thread #+srcname: bash-save-in-variable #+begin_src sh #!/bin/bash # so - Save Output. Saves output of command in OUT shell variable. OUT=`$*` echo $OUT #+end_src and another #+begin_quote .inputrc: "^[k": accept-line "^M": " | tee /tmp/h_lastcmd.out ^[k" .bash_profile: export __=/tmp/h_lastcmd.out If you try it, Alt-k will stand for the old Enter; use "command $__" to access the last output. Best, -- Herculano de Lima Einloft Neto #+end_quote ***** emacs-lisp will be a special case While it is possible for emacs-lisp to be run in a console type environment (see the =elim= function) it is *not* possible to run emacs-lisp in a different *session*. Meaning any variable set top level of the console environment will be set *everywhere* inside emacs. For this reason I think that it doesn't make any sense to worry about session support for emacs-lisp. *** Further thoughts on 'scripting' vs. functional approaches These are just thoughts, I don't know how sure I am about this. And again, perhaps I'm not saying anything very radical, just that it would be nice to have some options supporting things like receiving text output in the org buffer. I can see that you've already gone some way down the road towards the 'last value' approach, so sorry if my comments come rather late. I am concerned that we are not giving sufficient attention to stdout / the text that is returned by the interpreters. In contrast, many of our potential users will be accustomed to a 'scripting' approach, where they are outputting text at various points in the code block, not just at the end. I am leaning towards thinking that we should have 2 modes of evaluation: 'script' mode, and 'functional' mode. In script mode, evaluation of a code block would result in *all* text output from that code block appearing as output in the org buffer, presumably as an #+begin_example...#+end_example. There could be an :echo option controlling whether the input commands also appear in the output. [This is like Sweave]. In functional mode, the *result* of the code block is available as an elisp object, and may appear in the org buffer as an org table/string, via the mechanisms you have developed already. One thing I'm wondering about is whether, in script mode, there simply should not be a return value. Perhaps this is not so different from what exists: script mode would be new, and what exists currently would be functional mode. I think it's likely that, while code evaluation will be exciting to people, a large majority of our users in a large majority of their usage will not attempt to actually use the return value from a source code block in any meaningful way. In that case, it seems rather restrictive to only allow them to see output from the end of the code block. Instead I think the most accessible way to introduce org-babel to people, at least while they are learning it, is as an immensely powerful environment in which to embed their 'scripts', which now also allows them to 'run' their 'scripts'. Especially as such people are likely to be the least capable of the user-base, a possible design-rule would be to make the scripting style of usage easy (default?), perhaps requiring a special option to enable a functional style. Those who will use the functional style won't have a problem understanding what's going on, whereas the 'skript kiddies' might not even know the syntax for defining a function in their language of choice. And of course we can allow the user to set a variable in their .emacs controlling the preference, so that functional users are not inconveniennced by having to provide header args the whole time. Please don't get the impression that I am down-valuing the functional style of org-babel. I am constantly horrified at the messy 'scripts' that my colleagues produce in perl or R or whatever! Nevertheless that seems to be how a lot of people work. I think you were leaning towards the last-value approach because it offered the possibility of unified code supporting both the single evaluation environment and the functional style. If you agree with any of the above then perhaps it will impact upon this and mean that the code in the two branches has to differ a bit. In that case, functional mode could perhaps after all evaluate each code block in its own environment, thus (re)approaching 'true' functional programming (side-effects are hard to achieve). #+begin_src sh ls > files echo "There are `wc -l files` files in this directory" #+end_src *** even more thoughts on evaluation, results, models and options Thanks Dan, These comments are invaluable. What do you think about this as a new list of priorities/requirements for the execution of source-code blocks. - Sessions 1) we want the evaluation of the source code block to take place in a session which can persist state (variables, current directory, etc...). 2) source code blocks can specify their session with a header argument 3) each session should correspond to an Emacs comint buffer so that the user can drop into the session and experiment with live code evaluation. - Results 1) each source-code block generates some form of results which (as we have already implemented) is transfered into emacs-lisp after which it can be inserted into the org-mode buffer, or used by other source-code blocks 2) when the results are translated into emacs-lisp, forced to be interpreted as a scalar (dumping their raw values into the org-mode buffer), as a vector (which is often desirable with R code blocks), or interpreted on the fly (the default option). Note that this is very nearly currently implemented through the [[* DONE results-type header (vector/file)][results-type-header]]. 3) there should be *two* means of collecting results from the execution of a source code block. *Either* the value of the last statement of the source code block, or the collection of all that has been passed to STDOUT during the evaluation. **** header argument or return line (*header argument*) Rather than using a header argument to specify how the return value should be passed back, I'm leaning towards the use of a =#+RETURN= line inside the block. If such a line *is not present* then we default to using STDOUT to collect results, but if such a line *is present* then we use it's value as the results of the block. I think this will allow for the most elegant specification between functional and script execution. This also cleans up some issues of implementation and finding which statement is the last statement. Having given this more thought, I think a header argument is preferable. The =#+return:= line adds new complicating syntax for something that does little more than we would accomplish through the addition of a header argument. The only benefit being that we know where the final statement starts, which is not an issue in those languages which contain 'last value' operators. new header =:results= arguments - script :: explicitly states that we want to use STDOUT to initialize our results - return_last :: stdout is ignored instead the *value* of the final statement in the block is returned - echo :: means echo the contents of the source-code block along with the results (this implies the *script* =:results= argument as well) *** DONE rework evaluation lang-by-lang [4/4] This should include... - functional results working with the comint buffer - results headers - script :: return the output of STDOUT - write a macro which runs the first redirection, executes the body, then runs the second redirection - last :: return the value of the last statement - - sessions in comint buffers **** DONE Ruby [4/4] - [X] functional results working with comint - [X] script results - [X] ensure scalar/vector results args are taken into consideration - [X] ensure callable by other source block #+srcname: ruby-use-last-output #+begin_src ruby :results replace a = 2 b = 4 c = a + b [a, b, c, 78] #+end_src #+resname: ruby-use-last-output | 2 | 4 | 6 | 78 | #+srcname: task-call-use-last-output #+begin_src ruby :var last=ruby-use-last-output :results replace last.flatten.size + 1 #+end_src #+resname: task-call-use-last-output : 5 ***** ruby sessions #+srcname: first-ruby-session-task #+begin_src ruby :session schulte :results silent schulte = 27 #+end_src #+srcname: second-ruby-session-task #+begin_src ruby :session schulte :results silent schulte + 3 #+end_src #+srcname: without-the-right-session #+begin_src ruby :results silent schulte #+end_src **** DONE R [4/4] - [X] functional results working with comint - [X] script results - [X] ensure scalar/vector results args are taken into consideration - [X] ensure callable by other source block To redirect output to a file, you can use the =sink()= command. #+srcname: task_R_B #+begin_src R :results value vector silent a <- 9 b <- 10 b - a a + b #+end_src #+srcname: task-R-use-other-output #+begin_src R :var twoentyseven=task_R_B() :results replace value 83 twoentyseven + 9 #+end_src #+resname: task-R-use-other-output : 28 **** DONE Python [4/4] - [X] functional results working with comint - [X] script results - [X] ensure scalar/vector results args are taken into consideration - [X] ensure callable by other source block #+srcname: task-new-eval-for-python #+begin_src python :results silent output scalar 8 9 10 #+end_src #+srcname: task-use-new-eval #+begin_src python :var tasking=task-new-eval-for-python() :results replace tasking + 2 #+end_src #+resname: task-use-new-eval : 12 **** DONE Shells [4/4] - [X] functional results working with comint - [X] script results - [X] ensure scalar/vector results args are taken into consideration - [X] ensure callable by other source block #+srcname: task-shell-new-evaluation #+begin_src sh :results silent value scalar echo 'eric' date #+end_src #+srcname: task-call-other-shell #+begin_src sh :var other=task-shell-new-evaluation() :results replace scalar echo $other ' is the old date' #+end_src #+resname: task-call-other-shell : $ Fri Jun 12 13:08:37 PDT 2009 is the old date *** DONE implement a *session* header argument [4/4] =:session= header argument to override the default *session* buffer **** DONE ruby #+srcname: task-ruby-named-session #+begin_src ruby :session schulte :results replace schulte = :in_schulte #+end_src #+resname: task-ruby-named-session : :in_schulte #+srcname: another-in-schulte #+begin_src ruby :session schulte schulte #+end_src #+resname: another-in-schulte : :in_schulte : :in_schulte : :in_schulte **** DONE python #+srcname: python-session-task #+begin_src python :session what :results silent what = 98 #+end_src #+srcname: python-get-from-session #+begin_src python :session what :results replace what #+end_src #+resname: python-get-from-session : 98 **** DONE shell #+srcname: task-shell-sessions #+begin_src sh :session what WHAT='patton' #+end_src #+srcname: task-shell-sessions-what #+begin_src sh :session what :results replace echo $WHAT #+end_src #+resname: task-shell-sessions-what : patton **** DONE R #+srcname: task-R-session #+begin_src R :session what :results replace a <- 9 b <- 8 a + b #+end_src #+resname: task-R-session : 17 #+srcname: another-task-R-session #+begin_src R :session what :results replace a + b #+end_src *** DONE function to bring up inferior-process buffer [4/4] This should be callable from inside of a source-code block in an org-mode buffer. It should evaluate the header arguments, then bring up the inf-proc buffer using =pop-to-buffer=. For lack of a better place, lets add this to the `org-metadown-hook' hook. To give this a try, place the cursor on a source block with variables, (optionally git a prefix argument) then hold meta and press down. **** DONE ruby #+srcname: task-ruby-pop-to-session #+begin_src ruby :var num=9 :var another="something else" num.times{|n| puts another} #+end_src **** DONE python #+srcname: task-python-pop-to-session #+begin_src python :var num=9 :var another="something else" another * num #+end_src **** DONE R #+srcname: task-R-pop-to-session #+begin_src R :var a=9 :var b=8 a * b #+end_src **** DONE shell #+srcname: task-shell-pop-sessions #+begin_src sh :var NAME="eric" echo $NAME #+end_src *** DEFERRED function to dump last N lines from inf-proc buffer into the current source block Callable with a prefix argument to specify how many lines should be dumped into the source-code buffer. *** REJECTED comint notes Implementing comint integration in [[file:lisp/org-babel-comint.el][org-babel-comint.el]]. Need to have... - handling of outputs - split raw output from process by prompts - a ring of the outputs, buffer-local, `org-babel-comint-output-ring' - a switch for dumping all outputs to a buffer - inputting commands Lets drop all this language specific stuff, and just use org-babel-comint to split up our outputs, and return either the last value of an execution or the combination of values from the executions. **** comint filter functions : ;; comint-input-filter-functions hook process-in-a-buffer : ;; comint-output-filter-functions hook function modes. : ;; comint-preoutput-filter-functions hook : ;; comint-input-filter function ... #+srcname: obc-filter-ruby #+begin_src ruby :results last 1 2 3 4 5 #+end_src ** DONE Remove protective commas from # comments before evaluating org inserts protective commas in front of ## comments in language modes that use them. We need to remove them prior to sending code to the interpreter. #+srcname: testing-removal-of-protective-comas #+begin_src ruby ,# this one might break it?? :comma_protection #+end_src ** DONE pass multiple reference arguments into R Can we do this? I wasn't sure how to supply multiple 'var' header args. Just delete this if I'm being dense. This should be working, see the following example... #+srcname: two-arg-example #+begin_src R :var n=2 :var m=8 n + m #+end_src #+resname: two-arg-example : 10 ** DONE ensure that table ranges work when a table range is passed to org-babel as an argument, it should be interpreted as a vector. | 1 | 2 | simple | | 2 | 3 | Fixnum:1 | | 3 | 4 | Array:123456 | | 4 | 5 | | | 5 | 6 | | | 6 | 7 | | #+TBLFM: @1$3='(sbe simple-sbe-example (n 4))::@2$3='(sbe task-table-range (n @1$1..@6$1))::@3$3='(sbe task-table-range (n (@1$1..@6$1))) #+srcname: simple-sbe-example #+begin_src emacs-lisp "simple" #+end_src #+srcname: task-table-range #+begin_src ruby :var n=simple-sbe-example "#{n.class}:#{n}" #+end_src #+srcname: simple-results #+begin_src emacs-lisp :var n=task-table-range(n=(1 2 3)) n #+end_src #+resname: simple-results : Array:123 #+srcname: task-arr-referent #+begin_src ruby :var ar=(1 2 3) ar.size #+end_src #+resname: task-arr-referent : 3 ** DONE global variable indicating default to vector output how about an alist... =org-babel-default-header-args= this may already exist... just execute the following and all source blocks will default to vector output #+begin_src emacs-lisp (setq org-babel-default-header-args '((:results . "vector"))) #+end_src ** DONE name named results if source block is named currently this isn't happening although it should be #+srcname: test-naming-named-source-blocks #+begin_src emacs-lisp :namer #+end_src #+resname: test-naming-named-source-blocks : :namer ** DONE (simple caching) check for named results before source blocks see the TODO comment in [[file:lisp/org-babel-ref.el::TODO%20This%20should%20explicitly%20look%20for%20resname%20lines%20before][org-babel-ref.el#org-babel-ref-resolve-reference]] ** DONE set =:results silent= when eval with prefix argument #+begin_src emacs-lisp 'silentp #+end_src ** DONE results-type header (vector/file) [3/3] In response to a point in Dan's email. We should allow the user to force scalar or vector results. This could be done with a header argument, and the default behavior could be controlled through a configuration variable. #+srcname: task-trivial-vector #+begin_src ruby :results replace vector :scalar #+end_src #+resname: | ":scalar" | since it doesn't make sense to turn a vector into a scalar, lets just add a two values... - vector :: forces the results to be a vector (potentially 1 dimensional) - file :: this throws an error if the result isn't a string, and tries to treat it as a path to a file. I'm just going to cram all of these into the =:results= header argument. Then if we allow multiple header arguments it should work out, for example one possible header argument string could be =:results replace vector file=, which would *replace* any existing results forcing the results into an org-mode table, and interpreting any strings as file paths. *** DONE multiple =:results= headers #+srcname: multiple-result-headers #+begin_src ruby :results replace silent :schulte #+end_src #+resname: *** DONE file result types When inserting into an org-mode buffer create a link with the path being the value, and optionally the display being the =file-name-nondirectory= if it exists. #+srcname: task-file-result #+begin_src python :results replace file "something" #+end_src #+resname: [[something][something]] This will be useful because blocks like =ditaa= and =dot= can return the string path of their files, and can add =file= to their results header. *** DONE vector result types #+srcname: task-force-results #+begin_src emacs-lisp :results vector 8 #+end_src #+resname: | 8 | ** DONE results name In order to do this we will need to start naming our results. Since the source blocks are named with =#+srcname:= lines we can name results with =#+resname:= lines (if the source block has no name then no name is given to the =#+resname:= line on creation, otherwise the name of the source block is used). This will have the additional benefit of allowing results and source blocks to be located in different places in a buffer (and eventually in different buffers entirely). #+srcname: developing-resnames #+begin_src emacs-lisp :results silent 'schulte #+end_src Once source blocks are able to find their own =#+resname:= lines we then need to... #+srcname: sbe-w-new-results #+begin_src emacs-lisp :results replace (sbe "developing-resnames") #+end_src #+resname: : schulte *** TODO change the results insertion functions to use these lines *** TODO teach references to resolve =#+resname= lines. ** DONE org-babel tests org-babel [1/1] since we are accumulating this nice collection of source-code blocks in the sandbox section we should make use of them as unit tests. What's more, we should be able to actually use org-babel to run these tests. We would just need to cycle over every source code block under the sandbox, run it, and assert that the return value is equal to what we expect. I have the feeling that this should be possible using only org-babel functions with minimal or no additional elisp. It would be very cool for org-babel to be able to test itself. This is now done, see [[* Tests]]. *** DEFERRED org-babel assertions (may not be necessary) These could be used to make assertions about the results of a source-code block. If the assertion fails then the point could be moved to the block, and error messages and highlighting etc... could ensue ** DONE make C-c C-c work anywhere within source code block? This seems like it would be nice to me, but perhaps it would be inefficient or ugly in implementation? I suppose you could search forward, and if you find #+end_src before you find #+begin_src, then you're inside one. [DED] Agreed, I think inside of the =#+srcname: line= would be useful as well. #+srcname: testing-out-cc #+begin_src emacs-lisp 'schulte #+end_src ** DONE integration with org tables We should make it easy to call org-babel source blocks from org-mode table formulas. This is practical now that it is possible to pass arguments to org-babel source blocks. See the related [[* (sandbox) integration w/org tables][sandbox]] header for tests/examples. *** digging in org-table.el In the past [[file:~/src/org/lisp/org-table.el::org%20table%20el%20The%20table%20editor%20for%20Org%20mode][org-table.el]] has proven difficult to work with. Should be a hook in [[file:~/src/org/lisp/org-table.el::defun%20org%20table%20eval%20formula%20optional%20arg%20equation][org-table-eval-formula]]. Looks like I need to change this [[file:~/src/org/lisp/org-table.el::if%20lispp][if statement]] (line 2239) into a cond expression. ** DONE source blocks as functions Allow source code blocks to be called like functions, with arguments specified. We are already able to call a source-code block and assign it's return result to a variable. This would just add the ability to specify the values of the arguments to the source code block assuming any exist. For an example see When a variable appears in a header argument, how do we differentiate between it's value being a reference or a literal value? I guess this could work just like a programming language. If it's escaped or in quotes, then we count it as a literal, otherwise we try to look it up and evaluate it. ** DONE folding of code blocks? [2/2] [DED] In similar way to using outline-minor-mode for folding function bodies, can we fold code blocks? #+begin whatever statements are pretty ugly, and in any case when you're thinking about the overall game plan you don't necessarily want to see the code for each Step. *** DONE folding of source code block Sounds good, and wasn't too hard to implement. Code blocks should now be fold-able in the same manner as headlines (by pressing TAB on the first line). *** REJECTED folding of results So, lets do a three-stage tab cycle... First fold the src block, then fold the results, then unfold. There's no way to tell if the results are a table or not w/o actually executing the block which would be too expensive of an operation. ** DONE selective export of text, code, figures [DED] The org-babel buffer contains everything (code, headings and notes/prose describing what you're up to, textual/numeric/graphical code output, etc). However on export to html / LaTeX one might want to include only a subset of that content. For example you might want to create a presentation of what you've done which omits the code. [EMS] So I think this should be implemented as a property which can be set globally or on the outline header level (I need to review the mechanics of org-mode properties). And then as a source block header argument which will apply only to a specific source code block. A header argument of =:export= with values of - =code= :: just show the code in the source code block - =none= :: don't show the code or the results of the evaluation - =results= :: just show the results of the code evaluation (don't show the actual code) - =both= :: show both the source code, and the results this will be done in [[* (sandbox) selective export][(sandbox) selective export]]. ** DONE a header argument specifying silent evaluation (no output) This would be useful across all types of source block. Currently there is a =:replace t= option to control output, this could be generalized to an =:output= option which could take the following options (maybe more) - =t= :: this would be the default, and would simply insert the results after the source block - =replace= :: to replace any results which may already be there - =silent= :: this would inhibit any insertion of the results This is now implemented see the example in the [[* silent evaluation][sandbox]] ** DONE assign variables from tables in R This is now working (see [[* (sandbox table) R][(sandbox-table)-R]]). Although it's not that impressive until we are able to print table results from R. ** DONE insert 2-D R results as tables everything is working but R and shell *** DONE shells *** DONE R This has already been tackled by Dan in [[file:existing_tools/org-R.el::defconst%20org%20R%20write%20org%20table%20def][org-R:check-dimensions]]. The functions there should be useful in combination with [[http://cran.r-project.org/doc/manuals/R-data.html#Export-to-text-files][R-export-to-csv]] as a means of converting multidimensional R objects to emacs lisp. It may be as simple as first checking if the data is multidimensional, and then, if so using =write= to write the data out to a temporary file from which emacs can read the data in using =org-table-import=. Looking into this further, is seems that there is no such thing as a scalar in R [[http://tolstoy.newcastle.edu.au/R/help/03a/3733.html][R-scalar-vs-vector]] In that light I am not sure how to deal with trivial vectors (scalars) in R. I'm tempted to just treat them as vectors, but then that would lead to a proliferation of trivial 1-cell tables... ** DONE allow variable initialization from source blocks Currently it is possible to initialize a variable from an org-mode table with a block argument like =table=sandbox= (note that the variable doesn't have to named =table=) as in the following example #+TBLNAME: sandbox | 1 | 2 | 3 | | 4 | schulte | 6 | #+begin_src emacs-lisp :var table=sandbox :results replace (message (format "table = %S" table)) #+end_src : "table = ((1 2 3) (4 \"schulte\" 6))" It would be good to allow initialization of variables from the results of other source blocks in the same manner. This would probably require the addition of =#+SRCNAME: example= lines for the naming of source blocks, also the =table=sandbox= syntax may have to be expanded to specify whether the target is a source code block or a table (alternately we could just match the first one with the given name whether it's a table or a source code block). At least initially I'll try to implement this so that there is no need to specify whether the reference is to a table or a source-code block. That seems to be simpler both in terms of use and implementation. This is now working for emacs-lisp, ruby and python (and mixtures of the three) source blocks. See the examples in the [[* (sandbox) referencing other source blocks][sandbox]]. This is currently working only with emacs lisp as in the following example in the [[* emacs lisp source reference][emacs lisp source reference]]. ** TODO Add languages [0/5] I'm sure there are many more that aren't listed here. Please add them, and bubble any that you particularly care about up to the top. Any new language should be implemented in a org-babel-lang.el file. Follow the pattern set by [[file:lisp/org-babel-script.el][org-babel-script.el]], [[file:lisp/org-babel-shell.el][org-babel-shell.el]] and [[file:lisp/org-babel-R.el][org-babel-R.el]]. *** TODO perl This could probably be added to [[file:lisp/org-babel-script.el][org-babel-script.el]] *** TODO java *** TODO ditaa (see [[* file result types][file result types]]) *** TODO dot (see [[* file result types][file result types]]) *** TODO asymptote (see [[* file result types][file result types]]) * Bugs [17/26] ** TODO Allow source blocks to be recognised when #+ are not first characters on the line I think Carsten has recently altered the core so that #+ can have preceding whitespace, at least for literal/code examples. org-babel should support this. ** PROPOSED make :results replace the default? I'm tending to think that appending results to pre-existing results creates mess, and that the cleaner `replace' option should be the default. E.g. when a source block creates an image, we would want that to be updated, rather than have a new one be added. ** PROPOSED external shell execution can't isolate return values I have no idea how to do this as of yet. The result is that when shell functions are run w/o a session there is no difference between the =output= and =value= result arguments. ** TODO non-orgtbl formatted lists for example #+srcname: this-doesn't-match-orgtbl #+begin_src emacs-lisp :results replace '((:results . "replace")) #+end_src #+resname: this-doesn't-match-orgtbl ** TODO collapsing consecutive newlines in string output #+srcname: multi-line-string-output #+begin_src ruby :results replace "the first line ends here and this is the second one even a third" #+end_src #+resname: : the first line ends here : and this is the second one : return even a third ** TODO cursor movement when evaluating source blocks E.g. the pie chart example. Despite the save-window-excursion in org-babel-execute:R. (I never learned how to do this properly: org-R jumps all over the place...) ** TODO are the org-babel-trim s necessary? at the end of e.g. org-babel-R-evaluate, org-babel-python-evaluate, but not org-babel-ruby-evaluate ** TODO use new merge function [[file:lisp/org-babel-ref.el::t%20nil%20org%20combine%20plists%20args%20nil][here]] and [[file:lisp/org-babel.el::params%20org%20combine%20plists%20params%20third%20info][here]]? And at other occurrences of org-combine-plists? ** TODO #+lob: calls fail if reference has single character name *** This doesn't work #+lob: R-plot(data=X) #+tblname: X | 1 | 1 | | 2 | .5 | | 3 | .3333 | | 4 | .25 | | 5 | .2 | | 6 | .1666 | *** But this is OK #+tblname: XX | 1 | 1 | | 2 | .5 | | 3 | .3333 | | 4 | .25 | | 5 | .2 | | 6 | .1666 | #+lob: R-plot(data=XX) ** DEFERRED weird escaped characters in shell prompt break shell evaluation E.g. this doesn't work. Should the shell sessions set a sane prompt when they start up? Or is it a question of altering comint-prompt-regexp? Or altering org-babel regexps? #+begin_src sh black=30 ; red=31 ; green=32 ; yellow=33 ; blue=34 ; magenta=35 ; cyan=36 ; white=37 prompt_col=$red prompt_char='>' export PS1="\[\033[${prompt_col}m\]\w${prompt_char} \[\033[0m\]" #+end_src I just pushed a good amount of changes, could you see if your shell problems still exist? The problem's still there. Specifically, aIui, at [[file:lisp/langs/org-babel-sh.el::raw%20org%20babel%20comint%20with%20output%20buffer%20org%20babel%20sh%20eoe%20output%20nil%20insert%20full%20body%20comint%20send%20input%20nil%20t][this line]] of org-babel-sh.el, raw gets the value ("" " Sun Jun 14 19:26:24 EDT 2009\n" " org_babel_sh_eoe\n" " ") and therefore (member org-babel-sh-eoe-output ...) fails I think that `comint-prompt-regexp' needs to be altered to match the shell prompt. This shouldn't be too difficult to do by hand, using the `regexp-builder' command and should probably be part of the user's regular emacs init. I can't think of a way for us to set this automatically, and we are SOL without a regexp to match the prompt. ** DONE ruby evaluation not working under ubuntu emacs 23 With emacs 23.0.91.1 on ubuntu, for C-h f run-ruby I have the following, which seems to conflict with [[file:lisp/langs/org-babel-ruby.el::let%20session%20buffer%20save%20window%20excursion%20run%20ruby%20nil%20session%20current%20buffer][this line]] in org-babel-ruby.el. #+begin_example run-ruby is an interactive compiled Lisp function. (run-ruby cmd) Run an inferior Ruby process, input and output via buffer *ruby*. If there is a process already running in `*ruby*', switch to that buffer. With argument, allows you to edit the command line (default is value of `ruby-program-name'). Runs the hooks `inferior-ruby-mode-hook' (after the `comint-mode-hook' is run). (Type C-h m in the process buffer for a list of commands.) #+end_example So, I may have a non-standard inf-ruby.el. Here's my version of run-ruby. #+begin_example run-ruby is an interactive Lisp function in `inf-ruby.el'. (run-ruby &optional COMMAND NAME) Run an inferior Ruby process, input and output via buffer *ruby*. If there is a process already running in `*ruby*', switch to that buffer. With argument, allows you to edit the command line (default is value of `ruby-program-name'). Runs the hooks `inferior-ruby-mode-hook' (after the `comint-mode-hook' is run). (Type C-h m in the process buffer for a list of commands.) #+end_example It seems we could either bundle my version of inf-ruby.el (as it's the newest). Or we could change the use of `run-ruby' so that it is robust across multiple distributions. I think I'd prefer the former, unless the older version of inf-ruby is actually bundled with emacs, in which case maybe we should go out of our way to support it. Thoughts? I think for now I'll just include the latest [[file:util/inf-ruby.el][inf-ruby.el]] in the newly created utility directory. I doubt anyone would have a problem using the latest version of this file. ** DONE test failing forcing vector results with =test-forced-vector-results= ruby code block Note that this only seems to happen the *second* time the test table is evaluated #+srcname: bug-trivial-vector #+begin_src emacs-lisp :results vector silent 8 #+end_src #+srcname: bug-forced-vector-results #+begin_src ruby :var triv=test-trivial-vector :results silent triv.class.name #+end_src mysteriously this seems to be fixed... ** DONE defunct R sessions Sometimes an old R session will turn defunct, and newly inserted code will not be evaluated (leading to a hang). This seems to be fixed by using `inferior-ess-send-input' rather than `comint-send-input'. ** DONE ruby fails on first call to non-default session #+srcname: bug-new-session #+begin_src ruby :session is-new :patton #+end_src ** DONE when reading results from =#+resname= line Errors when trying to read from resname lines. #+resname: bug-in-resname : 8 #+srcname: bug-in-resname-reader #+begin_src emacs-lisp :var buggy=bug-in-resname() :results silent buggy #+end_src ** DONE R-code broke on "org-babel" rename #+srcname: bug-R-babels #+begin_src R 8 * 2 #+end_src ** DONE error on trivial R results So I know it's generally not a good idea to squash error without handling them, but in this case the error almost always means that there was no file contents to be read by =org-table-import=, so I think it's ok. #+srcname: bug-trivial-r1 #+begin_src R :results replace pie(c(1, 2, 3), labels = c(1, 2, 3)) #+end_src #+srcname: bug-trivial-r2 #+begin_src R :results replace 8 #+end_src #+resname: bug-trivial-r2 : 8 #+srcname: bug-trivial-r3 #+begin_src R :results replace c(1,2,3) #+end_src #+resname: bug-trivial-r3 | 1 | | 2 | | 3 | ** DONE ruby new variable creation (multi-line ruby blocks) Actually it looks like we were dropping all but the last line. #+srcname: multi-line-ruby-test #+begin_src ruby :var table=bug-numerical-table :results replace total = 0 table.each{|n| total += n} total/table.size #+end_src #+resname: : 2 ** DONE R code execution seems to choke on certain inputs Currently the R code seems to work on vertical (but not landscape) tables #+srcname: little-fake #+begin_src emacs-lisp "schulte" #+end_src #+begin_src R :var num=little-fake num #+end_src #+resname: : schulte #+srcname: set-debug-on-error #+begin_src emacs-lisp :results silent (setq debug-on-error t) #+end_src #+srcname: bug-numerical-table #+begin_src emacs-lisp :results silent '(1 2 3) #+end_src #+srcname: bug-R-number-evaluation #+begin_src R :var table=bug-numerical-table mean(mean(table)) #+end_src #+resname: : 2 #+tblname: bug-vert-table | 1 | | 2 | | 3 | #+srcname: bug-R-vertical-table #+begin_src R :var table=bug-vert-table :results silent mean(table) #+end_src ** DONE org bug/request: prevent certain org behaviour within code blocks E.g. [[]] gets recognised as a link (when there's text inside the brackets). This is bad for R code at least, and more generally could be argued to be inappropriate. Is it difficult to get org to ignore text in code blocks? [DED] I believe Carsten addressed this recently on the mailing list with the comment that it was indeed a difficult issue. I believe this may be one area where we could wait for an upstream (org-mode) fix. [Dan] Carsten has fixed this now in the core. ** DONE with :results replace, non-table output doesn't replace table output And vice versa. E.g. Try this first with table and then with len(table) [DED] #+begin_src python :var table=sandbox :results replace table #+end_src | 1 | 2 | 3 | | 4 | "schulte" | 6 | : 2 Yes, this is certainly a problem. I fear that if we begin replacing anything immediately following a source block (regardless of whether it matches the type of our current results) we may accidentally delete hand written portions of the user's org-mode buffer. I think that the best solution here would be to actually start labeling results with a line that looks something like... #+results: name This would have a couple of benefits... 1) we wouldn't have to worry about possibly deleting non-results (which is currently an issue) 2) we could reliably replace results even if there are different types 3) we could reference the results of a source-code block in variable definitions, which would be useful if for example we don't wish to re-run a source-block every time because it is long-running. Thoughts? If no-one objects, I believe I will implement the labeling of results. ** DONE extra quotes for nested string Well R appears to be reading the tables without issue... these *should* be quoted #+srcname: ls #+begin_src sh :results replace ls #+end_src | "COPYING" | | "README.markdown" | | "block" | | "examples.org" | | "existing_tools" | | "intro.org" | | "org-babel" | | "rorg.org" | | "test-export.html" | | "test-export.org" | #+srcname: test-quotes #+begin_src ruby :var tab=ls tab[1][0] #+end_src : README.markdown #+srcname: test-quotes #+begin_src R :var tab=ls as.matrix(tab[2,]) #+end_src : README.markdown ** DONE simple ruby arrays not working As an example eval the following. Adding a line to test #+tblname: simple-ruby-array | 3 | 4 | 5 | #+srcname: ruby-array-test #+begin_src ruby :var ar = simple-ruby-array :results silent ar.first.first #+end_src ** DONE space trailing language name fix regexp so it works when there's a space trailing the language name #+srcname: test-trailing-space #+begin_src ruby :schulte #+end_src ** DONE Args out of range error The following block resulted in the error below [DED]. It ran without error directly in the shell. #+begin_src sh cd ~/work/genopca for platf in ill aff ; do for pop in CEU YRI ASI ; do rm -f $platf/hapmap-genos-$pop-all $platf/hapmap-rs-all cat $platf/hapmap-genos-$pop-* > $platf/hapmap-genos-$pop-all cat $platf/hapmap-rs-* > $platf/hapmap-rs-all done done #+end_src executing source block with sh... finished executing source block string-equal: Args out of range: "", -1, 0 the error =string-equal: Args out of range: "", -1, 0= looks like what used to be output when the block returned an empty results string. This should be fixed in the current version, you should now see the following message =no result returned by source block=. ** DONE ruby arrays not recognized as such Something is wrong in [[file:lisp/org-babel-script.el]] related to the recognition of ruby arrays as such. #+begin_src ruby :results replace [1, 2, 3, 4] #+end_src | 1 | 2 | 3 | 4 | #+begin_src python :results replace [1, 2, 3, 4] #+end_src | 1 | 2 | 3 | 4 | * Tests Evaluate all the cells in this table for a comprehensive test of the org-babel functionality. *Note*: if you have customized =org-babel-default-header-args= then some of these tests may fail. #+TBLNAME: org-babel-tests | functionality | block | arg | expected | results | pass | |-------------------------+----------------------------+-----+-------------+-------------+------| | basic evaluation | | | | | pass | |-------------------------+----------------------------+-----+-------------+-------------+------| | emacs lisp | basic-elisp | | 5 | 5 | pass | | shell | basic-shell | | 6 | 6 | pass | | ruby | basic-ruby | | org-babel | org-babel | pass | | python | basic-python | | hello world | hello world | pass | | R | basic-R | | 13 | 13 | pass | |-------------------------+----------------------------+-----+-------------+-------------+------| | tables | | | | | pass | |-------------------------+----------------------------+-----+-------------+-------------+------| | emacs lisp | table-elisp | | 3 | 3 | pass | | ruby | table-ruby | | 1-2-3 | 1-2-3 | pass | | python | table-python | | 5 | 5 | pass | | R | table-R | | 3.5 | 3.5 | pass | |-------------------------+----------------------------+-----+-------------+-------------+------| | source block references | | | | | pass | |-------------------------+----------------------------+-----+-------------+-------------+------| | all languages | chained-ref-last | | Array | Array | pass | |-------------------------+----------------------------+-----+-------------+-------------+------| | source block functions | | | | | pass | |-------------------------+----------------------------+-----+-------------+-------------+------| | emacs lisp | defun-fibb | | fibbd | fibbd | pass | | run over | Fibonacci | 0 | 1 | 1 | pass | | a | Fibonacci | 1 | 1 | 1 | pass | | variety | Fibonacci | 2 | 2 | 2 | pass | | of | Fibonacci | 3 | 3 | 3 | pass | | different | Fibonacci | 4 | 5 | 5 | pass | | arguments | Fibonacci | 5 | 8 | 8 | pass | |-------------------------+----------------------------+-----+-------------+-------------+------| | bugs and tasks | | | | | pass | |-------------------------+----------------------------+-----+-------------+-------------+------| | | | simple ruby arrays | ruby-array-test | | 3 | 3 | | | R number evaluation | bug-R-number-evaluation | | 2 | | | | multi-line ruby blocks | multi-line-ruby-test | | 2 | | | | forcing vector results | test-forced-vector-results | | Array | | | |-------------------------+----------------------------+-----+-------------+-------------+------| | sessions | | | | | | |-------------------------+----------------------------+-----+-------------+-------------+------| | set ruby session | set-ruby-session-var | | :set | | | | get from ruby session | get-ruby-session-var | | 3 | | | | set python session | set-python-session-var | | set | | | | get from python session | get-python-session-var | | 4 | | | | set R session | set-R-session-var | | set | | | | get from R session | get-R-session-var | | 5 | | | #+TBLFM: $5='(if (= (length $3) 1) (progn (message (format "running %S" '(sbe $2 (n $3)))) (sbe $2 (n $3))) (sbe $2))::$6='(if (string= $4 $5) "pass" (format "expected %S but was %S" $4 $5)) ** basic tests #+srcname: basic-elisp #+begin_src emacs-lisp :results silent (+ 1 4) #+end_src #+srcname: basic-shell #+begin_src sh :results silent expr 1 + 5 #+end_src #+srcname: date-simple #+begin_src sh :results silent date #+end_src #+srcname: basic-ruby #+begin_src ruby :results silent "org-babel" #+end_src #+srcname: basic-python #+begin_src python :results silent 'hello world' #+end_src #+srcname: basic-R #+begin_src R :results silent b <- 9 b + 4 #+end_src ** read tables #+tblname: test-table | 1 | 2 | 3 | | 4 | 5 | 6 | #+srcname: table-elisp #+begin_src emacs-lisp :results silent :var table=test-table (length (car table)) #+end_src #+srcname: table-ruby #+begin_src ruby :results silent :var table=test-table table.first.join("-") #+end_src #+srcname: table-python #+begin_src python :var table=test-table table[1][1] #+end_src #+srcname: table-R #+begin_src R :var table=test-table mean(mean(table)) #+end_src ** references Lets pass a references through all of our languages... Lets start by reversing the table from the previous examples #+srcname: chained-ref-first #+begin_src python :var table = test-table table.reverse() table #+end_src #+resname: chained-ref-first | 4 | 5 | 6 | | 1 | 2 | 3 | Take the first part of the list #+srcname: chained-ref-second #+begin_src R :var table = chained-ref-first table[1] #+end_src #+resname: chained-ref-second | 4 | | 1 | Turn the numbers into string #+srcname: chained-ref-third #+begin_src emacs-lisp :var table = chained-ref-second (mapcar (lambda (el) (format "%S" el)) table) #+end_src #+resname: chained-ref-third | "(4)" | "(1)" | and Check that it is still a list #+srcname: chained-ref-last #+begin_src ruby :var table=chained-ref-third table.class.name #+end_src ** source blocks as functions #+srcname: defun-fibb #+begin_src emacs-lisp :results silent (defun fibbd (n) (if (< n 2) 1 (+ (fibbd (- n 1)) (fibbd (- n 2))))) #+end_src #+srcname: fibonacci #+begin_src emacs-lisp :results silent :var n=7 (fibbd n) #+end_src ** sbe tests (these don't seem to be working...) Testing the insertion of results into org-mode tables. #+srcname: multi-line-output #+begin_src ruby :results replace "the first line ends here and this is the second one even a third" #+end_src #+resname: : the first line ends here\n\n\n and this is the second one\n\neven a third #+srcname: multi-line-error #+begin_src ruby :results replace raise "oh nooooooooooo" #+end_src #+resname: : oh nooooooooooo | the first line ends here... | -:5: warning: parenthesize argument(s) for future version... | #+TBLFM: $1='(sbe "multi-line-output")::$2='(sbe "multi-line-error") ** forcing results types tests #+srcname: test-trivial-vector #+begin_src emacs-lisp :results vector silent 8 #+end_src #+srcname: test-forced-vector-results #+begin_src ruby :var triv=test-trivial-vector :results silent triv.class.name #+end_src ** sessions #+srcname: set-ruby-session-var #+begin_src ruby :session :results silent var = [1, 2, 3] :set #+end_src #+srcname: get-ruby-session-var #+begin_src ruby :session :results silent var.size #+end_src #+srcname: set-python-session-var #+begin_src python :session var=4 'set' #+end_src #+srcname: get-python-session-var #+begin_src python :session var #+end_src #+srcname: set-R-session-var #+begin_src R :session a <- 5 'set' #+end_src #+srcname: get-R-session-var #+begin_src R :session a #+end_src * Sandbox :PROPERTIES: :CUSTOM_ID: sandbox :END: To run these examples evaluate [[file:lisp/org-babel-init.el][org-babel-init.el]] ** org-babel.el beginning functionality #+begin_src sh :results replace date #+end_src #+resname: : Sun Jul 5 18:54:39 EDT 2009 #+begin_src ruby Time.now #+end_src #+resname: : Sun Jul 05 18:54:35 -0400 2009 #+begin_src python "Hello World" #+end_src #+resname: : Hello World ** org-babel-R #+begin_src R :results replace a <- 9 b <- 16 a + b #+end_src #+resname: : 25 #+begin_src R hist(rgamma(20,3,3)) #+end_src ** org-babel plays with tables Alright, this should demonstrate both the ability of org-babel to read tables into a lisp source code block, and to then convert the results of the source code block into an org table. It's using the classic "lisp is elegant" demonstration transpose function. To try this out... 1. evaluate [[file:lisp/org-babel-init.el]] to load org-babel and friends 2. evaluate the transpose definition =\C-c\\C-c= on the beginning of the source block 3. evaluate the next source code block, this should read in the table because of the =:var table=previous=, then transpose the table, and finally it should insert the transposed table into the buffer immediately following the block *** Emacs lisp #+begin_src emacs-lisp :results silent (defun transpose (table) (apply #'mapcar* #'list table)) #+end_src #+TBLNAME: sandbox | 1 | 2 | 3 | | 4 | schulte | 6 | #+begin_src emacs-lisp :var table=sandbox :results replace (transpose table) #+end_src #+begin_src emacs-lisp '(1 2 3 4 5) #+end_src #+resname: | 1 | 2 | 3 | 4 | 5 | *** Ruby and Python #+begin_src ruby :var table=sandbox :results replace table.first.join(" - ") #+end_src #+resname: : 1 - 2 - 3 #+begin_src python :var table=sandbox table[0] #+end_src #+resname: : [1, 2, 3] | 1 | 2 | 3 | #+begin_src ruby :var table=sandbox :results replace table #+end_src #+resname: : [[1, 2, 3], [4, "schulte", 6]] | 1 | 2 | 3 | | 4 | "schulte" | 6 | #+begin_src python :var table=sandbox :results replace len(table) #+end_src : 2 | "__add__" | "__class__" | "__contains__" | "__delattr__" | "__delitem__" | "__delslice__" | "__doc__" | "__eq__" | "__format__" | "__ge__" | "__getattribute__" | "__getitem__" | "__getslice__" | "__gt__" | "__hash__" | "__iadd__" | "__imul__" | "__init__" | "__iter__" | "__le__" | "__len__" | "__lt__" | "__mul__" | "__ne__" | "__new__" | "__reduce__" | "__reduce_ex__" | "__repr__" | "__reversed__" | "__rmul__" | "__setattr__" | "__setitem__" | "__setslice__" | "__sizeof__" | "__str__" | "__subclasshook__" | "append" | "count" | "extend" | "index" | "insert" | "pop" | "remove" | "reverse" | "sort" | *** (sandbox table) R #+TBLNAME: sandbox_r | 1 | 2 | 3 | | 4 | schulte | 6 | #+begin_src R :results replace x <- c(rnorm(10, mean=-3, sd=1), rnorm(10, mean=3, sd=1)) x #+end_src | -3.35473133869346 | | -2.45714878661 | | -3.32819924928633 | | -2.97310212756194 | | -2.09640758369576 | | -5.06054014378736 | | -2.20713700711221 | | -1.37618039712037 | | -1.95839385821742 | | -3.90407396475502 | | 2.51168071590226 | | 3.96753011570494 | | 3.31793212627865 | | 1.99829753972341 | | 4.00403686419829 | | 4.63723764452927 | | 3.94636744261313 | | 3.58355906547775 | | 3.01563442274226 | | 1.7634976849927 | #+begin_src R var tabel=sandbox_r :results replace tabel #+end_src | 1 | 2 | 3 | | 4 | "schulte" | 6 | *** shell Now shell commands are converted to tables using =org-table-import= and if these tables are non-trivial (i.e. have multiple elements) then they are imported as org-mode tables... #+begin_src sh :results replace ls -l #+end_src | "total" | 208 | "" | "" | "" | "" | "" | "" | | "-rw-r--r--" | 1 | "dan" | "dan" | 57 | 2009 | 15 | "block" | | "-rw-r--r--" | 1 | "dan" | "dan" | 35147 | 2009 | 15 | "COPYING" | | "-rw-r--r--" | 1 | "dan" | "dan" | 722 | 2009 | 18 | "examples.org" | | "drwxr-xr-x" | 4 | "dan" | "dan" | 4096 | 2009 | 19 | "existing_tools" | | "-rw-r--r--" | 1 | "dan" | "dan" | 2207 | 2009 | 14 | "intro.org" | | "drwxr-xr-x" | 2 | "dan" | "dan" | 4096 | 2009 | 18 | "org-babel" | | "-rw-r--r--" | 1 | "dan" | "dan" | 277 | 2009 | 20 | "README.markdown" | | "-rw-r--r--" | 1 | "dan" | "dan" | 11837 | 2009 | 18 | "rorg.html" | | "-rw-r--r--" | 1 | "dan" | "dan" | 61829 | 2009 | 19 | "#rorg.org#" | | "-rw-r--r--" | 1 | "dan" | "dan" | 60190 | 2009 | 19 | "rorg.org" | | "-rw-r--r--" | 1 | "dan" | "dan" | 972 | 2009 | 11 | "test-export.org" | ** silent evaluation #+begin_src ruby :im_the_results #+end_src : :im_the_results #+begin_src ruby :results silent :im_the_results #+end_src #+begin_src ruby :results replace :im_the_results_ #+end_src : :im_the_results_ ** (sandbox) referencing other source blocks Doing this in emacs-lisp first because it's trivial to convert emacs-lisp results to and from emacs-lisp. *** emacs lisp source reference This first example performs a calculation in the first source block named =top=, the results of this calculation are then saved into the variable =first= by the header argument =:var first=top=, and it is used in the calculations of the second source block. #+SRCNAME: top #+begin_src emacs-lisp (+ 4 2) #+end_src #+begin_src emacs-lisp :var first=top :results replace (* first 3) #+end_src : 18 This example is the same as the previous only the variable being passed through is a table rather than a number. #+begin_src emacs-lisp :results silent (defun transpose (table) (apply #'mapcar* #'list table)) #+end_src #+TBLNAME: top_table | 1 | 2 | 3 | | 4 | schulte | 6 | #+SRCNAME: second_src_example #+begin_src emacs-lisp :var table=top_table (transpose table) #+end_src #+begin_src emacs-lisp :var table=second_src_example :results replace (transpose table) #+end_src | 1 | 2 | 3 | | 4 | "schulte" | 6 | *** ruby python Now working for ruby #+srcname: start #+begin_src ruby 89 #+end_src #+begin_src ruby :var other=start :results replace 2 * other #+end_src and for python #+SRCNAME: start_two #+begin_src python 98 #+end_src #+begin_src python :var another=start_two :results replace another*3 #+end_src *** mixed languages Since all variables are converted into Emacs Lisp it is no problem to reference variables specified in another language. #+SRCNAME: ruby-block #+begin_src ruby 2 #+end_src #+SRCNAME: lisp_block #+begin_src emacs-lisp :var ruby-variable=ruby-block (* ruby-variable 8) #+end_src #+begin_src python :var lisp_var=lisp_block lisp_var + 4 #+end_src : 20 *** R #+srcname: first_r #+begin_src R :results replace a <- 9 a #+end_src : 9 #+begin_src R :var other=first_r :results replace other + 2 #+end_src : 11 ** (sandbox) selective export For exportation tests and examples see (including exportation of inline source code blocks) [[file:test-export.org]] ** (sandbox) source blocks as functions #+srcname: default #+begin_src emacs-lisp :results silent 5 #+end_src #+srcname: triple #+begin_src emacs-lisp :var n=default :results replace (* 3 n) #+end_src : 15 #+begin_src emacs-lisp :var result=triple(n=3, m=98) :results replace result #+end_src : 294 The following just demonstrates the ability to assign variables to literal values, which was not implemented until recently. #+begin_src ruby :var num="eric" :results replace num+" schulte " #+end_src : "eric schulte " ** (sandbox) inline source blocks This is an inline source code block src_ruby{1 + 6}. And another source block with text output src_emacs-lisp{"eric"}. This is an inline source code block with header arguments. src_ruby[:var n=fibbd( n = 0 )]{n} ** (sandbox) integration w/org tables #+begin_src emacs-lisp :results silent (defun fibbd (n) (if (< n 2) 1 (+ (fibbd (- n 1)) (fibbd (- n 2))))) #+end_src #+srcname: fibbd #+begin_src emacs-lisp :var n=4 :results silent (fibbd n) #+end_src #+begin_src emacs-lisp :results silent (mapcar #'fibbd '(0 1 2 3 4 5 6 7 8)) #+end_src Something is not working here. The function `sbe ' works fine when called from outside of the table (see the source block below), but produces an error when called from inside the table. I think there must be some narrowing going on during intra-table emacs-lisp evaluation. | original | fibbd | |----------+-------| | 0 | 1 | | 1 | 1 | | 2 | 2 | | 3 | 3 | | 4 | 5 | | 5 | 8 | | 6 | 13 | | 7 | 21 | | 8 | 34 | | 9 | 55 | #+TBLFM: $2='(sbe "fibbd" (n $1)) silent-result #+begin_src emacs-lisp :results silent (sbe 'fibbd (n "8")) #+end_src * Buffer Dictionary LocalWords: DBlocks dblocks org-babel el eric fontification