Re: [Patch] contrib/mklog.py: Improve PR handling

[Martin Sebor <msebor@gmail.com>]

On 6/18/21 5:25 AM, Tobias Burnus wrote:
> On 18.06.21 13:10, Jonathan Wakely wrote:
> 
>> On Fri, 18 Jun 2021 at 12:05, Tobias Burnus wrote:
>>> PR c++/12394 - internal compiler error: in write_type, at 
>>> cp/mangle.c:1517
>>> PR fortran/100123 - -ftree-fre gives incorrect result in subroutine 
>>> with array declared as length 1
>>>           PR c++/12394
>>>           PR fortran/100123
>> Now that we put these PR cmpt/nnnn lines before the xxx/ChangeLog
>> entries, is there any reason to indent them with a TAB?
> 
> My understanding with the PR before vs. after the xxx/ChangeLog is
> that if you put them before, they apply to all changelog entries
> and if you put them afterward, it only only applies to the those
> ChangeLog files for which they have been mentioned.
> 
> (But I also might have misread the code/were misguided by the
> var names.)
> 
> And when manually creating the changelog from scratch, they still
> often end up after the ChangeLog line.
> 
> But otherwise, I am not attached to the tab. Just one thing:
> It might be slightly inconsistent to require the tab after the
> xxx/ChangeLog line but not before that line. And I think a
> tab should be used after the xxx/ChangeLog line.
> 
>> If we didn't require the TAB before them, then the lines added by -p
>> would serve the same purpose, and we wouldn't need to repeat them
>>
>> So instead of requiring "^\tPR .*/\d+$" for the PR entry, require
>> "^\t?PR ([^/]+)/\d+"
> I think we should require either ($|\s) after the PR number but
> otherwise, it works for me.
> 
> I am sure that Martin Sebor will like this change as he currently
> writes changelogs (see in this thread or) like the following one.

I do like adding the bug title and extracting the component from
Bugzilla.  IIRC, Martin Liska implemented -p in mklog.py on my
request (although I just discovered the option recently). I was
going to propose something similar to your solution:
   https://gcc.gnu.org/pipermail/gcc/2021-June/236460.html
but I'm happy to let you run with it.

Thanks
Martin

> 
> In that sense, there would be a precedent in terms of actual usage.
> 
> ----------
> Teach compute_objsize about placement new [PR100876].
> 
> Resolves:
> PR c++/100876 - -Wmismatched-new-delete should understand placement new 
> when it's not inlined
> 
> gcc/ChangeLog:
> 
>         PR c++/100876
> ----------
> 
> Tobias
> 
> PS: My feeling is that you like the proposed mklog patch :-)
> 
> -----------------
> Mentor Graphics (Deutschland) GmbH, Arnulfstrasse 201, 80634 München 
> Registergericht München HRB 106955, Geschäftsführer: Thomas Heurung, 
> Frank Thürauf