Re: [PATCH] Babel evaluation: location and timing information

[Ihor Radchenko <yantar92@gmail.com>]

Timothy <orgmode@tec.tecosaur.net> writes:

> Hi Ihor,
>
>>> -	    (message “executing %s code block%s…”
>>> +	    (message “executing %s %s %s…”
>>>  		     (capitalize lang)
>>> +                     (pcase (org-element-type (org-element-at-point))
>>> +                       (’src-block “code block”)
>>> +                       (’babel-call “call”)
>>> +                       (’paragraph “inline code block”))
>>
>> This will not work, for example, when inline src block is located inside
>> a headline. One can think of various other non-paragraph scenarios as well.
>
> Good point. Does assuming anything other than a src block or babel call is an
> inline src block sound reasonable?

Yes. All other elements (from org-element-all-elements) are
non-source-blocks. It will be a good idea to spell out this fact right
in the code (in comment).

>> Also, even though org-element-at-point should be caching recent calls,
>> I’d try to test the performance before/after the patch on large number
>> of src blocks (like in your config). org-element-at-point can add a fair
>> bit of CPU load in some scenarios where we have fallback to the full
>> O(Log N) AVL-tree lookup.
>
> I’ll see about giving this a shot at some point. Since parsing would be required
> before this point I assumed the document structure would be cached and this
> should be fairly cheap. Would `org-element-at-point-no-context' be better
> potentially?

No, `org-element-at-point-no-context' will not be better. It is only
useful in very specific situations when buffer changes are large,
frequent, affect large portions of cache, and parsing from the parent
heading is faster than updating the change all the way to the
top-parent org-data element.

It will be better to re-use the element queried during src block
execution.

>>> +                (let ((time-info
>>> +                       (if (and exec-time (> (float-time exec-time) 0.05))
>>> +                           (format “ (took %.1fs)” (float-time exec-time))
>>
>> Why 0.05??
>
> Because that’s half the resolution at which the time is displayed at. The idea
> is to only show the time when it will show something greater than zero.

Makes sense. Could you please add this comment to the code?

-- 
Ihor Radchenko,
Org mode contributor,
Learn more about Org mode at https://orgmode.org/.
Support Org development at https://liberapay.com/org-mode,
or support my work at https://liberapay.com/yantar92