org-mode/org-babel.org

2847 lines
92 KiB
Org Mode
Raw Normal View History

2009-05-24 17:10:03 -04:00
#+TITLE: org-babel --- facilitating communication between programming languages and people
2009-06-12 16:09:20 -04:00
#+SEQ_TODO: TODO PROPOSED | DONE DEFERRED REJECTED
#+OPTIONS: H:3 num:nil toc:t
#+STARTUP: oddeven hideblocks
* Introduction
Org-Babel enables *communication* between programming languages and
between people.
Org-Babel provides:
- communication between programs :: Data passes seamlessly between
different programming languages, text, and tables.
- communication between people :: Data and calculations are embedded
in the same document as notes explanations and reports.
** communication between programs
2009-05-25 21:20:32 -04:00
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
2009-05-25 18:13:56 -04:00
*** operations in/on tables
#+tblname: grades-table
2009-05-24 18:40:54 -04:00
| student | grade | letter |
|---------+-------+--------|
| 1 | 99 | A |
| 2 | 59 | F |
| 3 | 75 | C |
| 4 | 15 | F |
| 5 | 7 | F |
| 6 | 13 | F |
2009-05-24 18:40:54 -04:00
#+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
2009-05-25 18:13:56 -04:00
** communication between people
2009-05-25 21:20:32 -04:00
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.
2009-06-15 15:47:10 -04:00
* Tasks [27/42]
** TODO support for working with =*Org Edit Src Example*= buffers [1/4]
*** 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.
*** TODO some possible requests/proposed changes for Carsten [2/3]
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
**** TODO 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.
**** DEFERRED a hook called when the src edit buffer is created
This should be implemented in the org-mode core
*** 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]
*** 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]
** 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
** 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.
** TODO share org-babel
how should we share org-babel?
- post to org-mode and ess mailing lists
- create a org-babel page on worg
- 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]].
** 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))))
2009-06-12 16:09:20 -04:00
#+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.
2009-06-12 16:09:20 -04:00
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]
2009-06-12 16:09:20 -04:00
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=
2009-06-12 18:31:52 -04:00
** 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
2009-06-12 18:31:52 -04:00
#+end_src
*** DEFERRED use =C-g= keyboard-quit to push processing into the background
This may be possible using the `run-with-timer' command.
2009-06-12 18:31:52 -04:00
I have no idea how this could work...
#+srcname: long-runner-ruby
#+begin_src ruby :results silent
sleep(10)
:patton_is_an_grumpy
2009-06-12 18:31:52 -04:00
#+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.
2009-06-14 14:48:11 -04:00
[[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
2009-06-15 15:47:10 -04:00
** 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.
2009-06-15 14:49:06 -04:00
** 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
2009-05-23 11:12:12 -04:00
and another
2009-05-23 11:12:12 -04:00
#+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
2009-06-13 19:01:06 -04:00
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.
2009-06-13 19:01:06 -04:00
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.
2009-05-09 21:51:37 -04:00
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].
2009-05-09 21:51:37 -04:00
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.
2009-05-09 21:51:37 -04:00
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.
2009-05-09 21:51:37 -04:00
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).
2009-05-09 21:51:37 -04:00
#+begin_src sh
ls > files
echo "There are `wc -l files` files in this directory"
2009-05-09 21:51:37 -04:00
#+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*)
2009-05-09 21:51:37 -04:00
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.
2009-05-09 21:51:37 -04:00
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.
2009-05-09 21:51:37 -04:00
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)
2009-05-09 21:51:37 -04:00
*** 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
2009-05-09 21:51:37 -04:00
**** 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=.
2009-06-14 21:48:43 -04:00
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
2009-05-24 13:35:13 -04:00
** DONE ensure that table ranges work
when a table range is passed to org-babel as an argument, it should be
2009-05-24 13:35:13 -04:00
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.
2009-05-09 21:33:08 -04:00
** 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.
2009-04-23 18:59:42 -04:00
** DONE selective export of text, code, figures
[DED] The org-babel buffer contains everything (code, headings and
2009-04-23 18:59:42 -04:00
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]].
2009-04-05 19:27:14 -04:00
** 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.
2009-04-05 19:27:14 -04:00
** DONE insert 2-D R results as tables
2009-04-03 19:50:57 -04:00
everything is working but R and shell
2009-04-05 19:27:14 -04:00
*** DONE shells
2009-04-03 19:50:57 -04:00
*** DONE R
2009-04-03 19:50:57 -04:00
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
2009-04-03 19:50:57 -04:00
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...
2009-04-05 19:27:14 -04:00
** DONE allow variable initialization from source blocks
2009-04-01 18:59:33 -04:00
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]].
2009-05-22 18:11:53 -04:00
** 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]].
2009-05-22 18:11:53 -04:00
*** TODO perl
This could probably be added to [[file:lisp/org-babel-script.el][org-babel-script.el]]
2009-05-22 18:11:53 -04:00
*** 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]])
2009-06-15 14:29:25 -04:00
* Bugs [16/21]
** 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 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
2009-06-14 15:22:11 -04:00
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" " ")
2009-06-14 21:48:43 -04:00
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.
** 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
2009-05-22 09:56:34 -04:00
** 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...)
** 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
2009-05-23 17:22:46 -04:00
** 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)
2009-05-23 17:22:46 -04:00
#+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.
2009-05-22 09:56:34 -04:00
#+srcname: multi-line-ruby-test
2009-05-22 09:56:34 -04:00
#+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
: 11
: 11
: 11
: schulte
: 9
: 9
: 11
#+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 :results replace
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
** DEFERRED 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.
** DONE with :results replace, non-table output doesn't replace table output
2009-05-14 20:35:29 -04:00
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
2009-06-14 15:09:36 -04:00
#+tblname: simple-ruby-array
| 3 | 4 | 5 |
#+srcname: ruby-array-test
#+begin_src ruby :var ar = simple-ruby-array :results silent
2009-06-14 15:09:36 -04:00
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
2009-04-05 22:22:50 -04:00
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 |
2009-04-23 18:59:42 -04:00
* 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
2009-06-15 15:30:36 -04:00
| 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 | pass |
| R number evaluation | bug-R-number-evaluation | | 2 | 2 | pass |
| multi-line ruby blocks | multi-line-ruby-test | | 2 | 2 | pass |
| forcing vector results | test-forced-vector-results | | Array | Array | pass |
|-------------------------+----------------------------+-----+-------------+-------------+------|
| sessions | | | | | pass |
|-------------------------+----------------------------+-----+-------------+-------------+------|
| set ruby session | set-ruby-session-var | | :set | :set | pass |
| get from ruby session | get-ruby-session-var | | 3 | 3 | pass |
| set python session | set-python-session-var | | set | set | pass |
| get from python session | get-python-session-var | | 4 | 4 | pass |
| set R session | set-R-session-var | | set | set | pass |
| get from R session | get-R-session-var | | 5 | 5 | pass |
#+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
2009-06-13 17:49:57 -04:00
#+srcname: date-simple
#+begin_src sh :results silent
2009-06-13 17:49:57 -04:00
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
2009-04-01 18:59:33 -04:00
#+begin_src sh :results replace
date
#+end_src
2009-05-14 20:35:29 -04:00
: Thu May 14 18:52:25 EDT 2009
#+begin_src ruby
2009-05-09 21:28:08 -04:00
Time.now
#+end_src
2009-05-14 20:35:29 -04:00
: Thu May 14 18:59:09 -0400 2009
2009-05-09 21:28:08 -04:00
#+begin_src python
2009-05-09 21:28:08 -04:00
"Hello World"
#+end_src
2009-05-09 21:28:08 -04:00
: Hello World
** org-babel-R
2009-04-01 18:59:33 -04:00
#+begin_src R :results replace
a <- 9
b <- 16
a + b
#+end_src
: 25
2009-03-31 14:54:19 -04:00
#+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
2009-05-14 20:35:29 -04:00
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
2009-05-09 21:28:08 -04:00
#+begin_src emacs-lisp :results silent
(defun transpose (table)
(apply #'mapcar* #'list table))
#+end_src
2009-05-14 20:35:29 -04:00
#+TBLNAME: sandbox
| 1 | 2 | 3 |
| 4 | schulte | 6 |
2009-04-01 18:59:33 -04:00
#+begin_src emacs-lisp :var table=sandbox :results replace
(transpose table)
#+end_src
2009-05-09 21:28:08 -04:00
#+begin_src emacs-lisp
'(1 2 3 4 5)
#+end_src
| 1 | 2 | 3 | 4 | 5 |
*** Ruby and Python
2009-04-01 18:59:33 -04:00
#+begin_src ruby :var table=sandbox :results replace
table.first.join(" - ")
#+end_src
: "1 - 2 - 3"
2009-04-01 18:59:33 -04:00
#+begin_src python :var table=sandbox :results replace
table[0]
#+end_src
| 1 | 2 | 3 |
2009-04-01 18:59:33 -04:00
#+begin_src ruby :var table=sandbox :results replace
table
#+end_src
| 1 | 2 | 3 |
| 4 | "schulte" | 6 |
2009-04-01 18:59:33 -04:00
#+begin_src python :var table=sandbox :results replace
2009-05-14 20:35:29 -04:00
len(table)
#+end_src
2009-05-14 20:35:29 -04:00
: 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 |
2009-04-01 18:59:33 -04:00
#+begin_src R :results replace
x <- c(rnorm(10, mean=-3, sd=1), rnorm(10, mean=3, sd=1))
x
#+end_src
2009-05-14 20:35:29 -04:00
| -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...
2009-05-09 21:28:08 -04:00
#+begin_src sh :results replace
ls -l
#+end_src
2009-05-14 20:35:29 -04:00
| "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" |
2009-05-14 20:35:29 -04:00
| "-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
2009-05-14 20:35:29 -04:00
: :im_the_results
#+begin_src ruby :results silent
:im_the_results
#+end_src
#+begin_src ruby :results replace
2009-05-14 20:35:29 -04:00
:im_the_results_
#+end_src
2009-05-14 20:35:29 -04:00
: :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
2009-05-14 20:35:29 -04:00
: 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
2009-05-14 20:35:29 -04:00
| 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
2009-05-09 21:28:08 -04:00
: 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]]
2009-04-22 19:18:42 -04:00
2009-04-24 19:48:41 -04:00
** (sandbox) source blocks as functions
#+srcname: default
2009-05-09 21:28:08 -04:00
#+begin_src emacs-lisp :results silent
2009-04-24 19:48:41 -04:00
5
#+end_src
#+srcname: triple
#+begin_src emacs-lisp :var n=default :results replace
(* 3 n)
#+end_src
: 15
2009-05-14 20:35:29 -04:00
#+begin_src emacs-lisp :var result=triple(n=3, m=98) :results replace
2009-04-24 19:48:41 -04:00
result
#+end_src
2009-05-14 20:35:29 -04:00
: 294
2009-05-09 21:28:08 -04:00
The following just demonstrates the ability to assign variables to
literal values, which was not implemented until recently.
2009-05-09 21:28:08 -04:00
#+begin_src ruby :var num="eric" :results replace
2009-05-14 20:35:29 -04:00
num+" schulte "
#+end_src
2009-05-14 20:35:29 -04:00
: "eric schulte "
2009-04-24 19:48:41 -04:00
2009-05-09 21:51:37 -04:00
** (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
2009-05-14 20:35:29 -04:00
#+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
2009-05-09 21:51:37 -04:00
* Buffer Dictionary
LocalWords: DBlocks dblocks org-babel el eric fontification