[Orgmode] Re: [BABEL] tangling with leaving non-source code lines as empty lines?

[Eric Schulte]

Hi Rainer,

Dr Rainer M Krug <address@hidden> writes:

> On 13/07/2010 02:46, Eric Schulte wrote:
>> Hi Rainer,
>> 
>> With the tangling comments generated by the latest Org-mode, the
>> following function should be able to jump from any code in a tangled
>> code file back to the relevant block in the original Org-mode file.
>
> Wow - that was quick.
>
> Thanks.
>
>> 
>> --8<---------------cut here---------------start------------->8---
>> (defun org-babel-tangle-jump-to-org ()
>>   "Jump from a tangled code file to the related Org-mode file."
>>   (interactive)
>>   (let ((mid (point))
>>      target-buffer target-char
>>      start end link path block-name)
>>     (save-window-excursion
>>       (save-excursion
>>      (unless (and (re-search-backward org-bracket-link-analytic-regexp nil t)
>>                   (setq start (point))
>>                   (setq link (match-string 0))
>>                   (setq path (match-string 3))
>>                   (setq block-name (match-string 5))
>>                   (re-search-forward (concat " " (regexp-quote block-name)
>>                                              " ends here[\n\r]") nil t)
>>                   (setq end (point))
>>                   (< start mid) (< mid end))
>>        (error "not in tangled code")))
>>       (when (string-match "::" path)
>>      (setq path (substring path 0 (match-beginning 0))))
>>       (find-file path) (setq target-buffer (current-buffer))
>>       (goto-char start) (org-open-link-from-string link)
>>       (if (string-match "[^ \t\n\r]:\\([[:digit:]]+\\)" block-name)
>>        (org-babel-next-src-block
>>         (string-to-int (match-string 1 block-name)))
>>      (org-babel-goto-named-src-block block-name))
>>       (setq target-char (point)))
>>     (pop-to-buffer target-buffer)
>>     (goto-char target-char)))
>> --8<---------------cut here---------------end--------------->8---
>> 
>
> I tried it out, and realised that one needs to name the blocks
> (#+srcname:) to be able to use it. Then it jumps to the code block in
> the org file.
>

Have you tried this on un-named code blocks (with the latest Org-mode
from git)?  It should work for those as well -- it parses the name, and
uses org-babel-next-src-block to jump down the appropriate number of
blocks.

>
> Two more things (As pointed out in the other email) would be nice:
>
> 1) it would be nice to jump to the line of code in the Org-mode file
> which corresponds to the line in the code file
>

Agreed, I was planning on implementing this, it shouldn't be difficult.

>
> 2) It would be brilliant, if one could call the function from the
> Org-file buffer and enter the line number and then jump to the block /
> line in the block.
>

That's a great suggestions, I will certainly look at folding it in.

>
> In addition:
> 3) it would be very useful, if this function could be used with
> non-named source blocks.
>

As I said above, I /believe/ that it is usable as such now, however if
you could come up with an example where this fails, please do let me
know.

>
> But that function is already quite usefull.
>

Great, I'll post any future improvements I make to this thread, and
please do let me know if you run across errant behavior or new
potentially useful behavior.

Cheers -- Eric

>
> Thanks,
>
> Rainer
>
>
>> This is too large of a block of untested code to push into Org-mode now,
>> however please give it a try, and let me know if you think this could be
>> generally useful.  Hopefully after the feature-freeze we can fold
>> something like this into Babel to ease navigation between Org files and
>> their tangled offspring.
>> 
>> Cheers -- Eric
>> 
>> "Eric Schulte" <address@hidden> writes:
>> 
>>> Hi Rainer,
>>>
>>> If I'm understand you correctly you want the absolute position (by LOC)
>>> of the lines of R code to be the same in both the original org file and
>>> the tangled file or R code.  I don't think this is possible.  It is
>>> possible to tangle code blocks in any order (not just the order in which
>>> they appear in the original org file), so there are many valid tangling
>>> scenarios in which this feature could not work.
>>>
>>> Rather, I think that it may make sense to have a function for jumping
>>> back and forth between tangled code blocks and the related Org file.  In
>>> that case the R errors could be followed to the tangled R code file,
>>> from which you could easily jump to the relevant line in the relevant
>>> code block in the Org file.  If you are tangling with comments, then it
>>> should be possible to have a function called from a line in a tangled
>>> code file which when called
>>>
>>> 1) finds it's enclosing comments
>>> 2) remembers it's offset from the comments (which would then be it's
>>>    offset in the code block in the Org file)
>>> 3) read the comment to learn which code block in which Org file it's
>>>    tangled from
>>> 4) jump to the relevant line, in the relevant block, in the relevant
>>>    file
>>>
>>> Similarly when called form within a code block in an Org file the
>>> function could
>>>
>>> 1) read it's header argument to find the relevant tangled code file
>>> 2) jump to that file
>>> 3) use the comments in that file to move to the appropriate section of
>>>    code and related line
>>>
>>> I'll think about such a function, and if it makes sense to implement it
>>> apart from a more general "activate org-mode links in comments" minor
>>> mode.  Any ideas or suggestions would be welcome!
>>>
>>> Thanks -- Eric
>>>
>>> Rainer M Krug <address@hidden> writes:
>>>
>>>> Hi Eric,
>>>>
>>>> would it be possible, when tangling, to leave lines without source code in
>>>> the resulting code file as empty lines? The reasoning would be that error
>>>> messages (at least in R) give the line in which the error occurred. If the
>>>> line numbers in which the code sits would be preserved (by leaving empty
>>>> lines where no code is in the .org file), debugging would be much easier.
>>>>
>>>> By the way: I am using the after tangle hook and it works absolutely
>>>> perfectly.
>>>>
>>>> Thanks and cheers,
>>>>
>>>> Rainer