Re: [NonGNU ELPA] New package: eldoc-diffstat

[Philip Kaludercic <philipk@posteo.net>]

Johann Klähn <johann@jklaehn.de> writes:

> Thanks for your comments, everyone!

[...]

> On Sat, Dec 14, 2024 at 10:11 +0000, Philip Kaludercic wrote:
>>    '((Git "git" "--no-pager" "show" "--color=always"
>> -         "--format=format:%an <%ae>, %aD:%n%s" "--stat=80")
>> +	 "--format=format:%an <%ae>, %aD:%n%s" "--stat=80")
>
> Some lines look like spurious whitespace changes?  I hope I picked out
> all functional changes.

That is true, my apologies.

>> +  ;; Is it an issue that there is the slight possibility of a race
>> +  ;; condition here?
>
> I happened to have the same thought yesterday, also w.r.t. the
> sentinel function.

As far as I see, there is no way of avoiding this without extending
`make-process' to accept a process plist (which wouldn't be backwards
compatible).  But if you don't insist on using the process plist, you
could capture the revision info in a closure and pass it on to the
sentinel.

>>                                         The rewriting of
>> `eldoc-diffstat--format-output-buffer' might be controversial, but I
>> feel that using a regular expression to destruct the buffer feels more
>> robust.
>
> Yes, probably.  Though I decided to keep the original output if the
> regex fails to match, as it might help users figure out what's happening.

That's fine as well of course!