Re: lo-commit-stat: why lowercase the first letter?

Korrawit Pruegsanusak píše v Ne 20. 05. 2012 v 11:28 +0700:
> Hello Petr, all,
>
> Seeing a fix in wiki [1], IIUC this is generated by lo-commit-stat,
> which is intended to lowercase the first letter [2].
>
> By git annotate, this part of code is there since the start, so I
> would like to ask:
> * why lowercase the first letter?

Every developer uses another style for the summary. I did not want to
make complicated rules and annoyed developers with "Hey, do right commit
messages" mails. So, I tried to make the commit summary more consistent
by the script. It is supposed to do:

        + use lower case for the first letter[*]
        + move bugzilla numbers at the end of the line[**]
        + put bugzilla numbers into brackets to make it better visually
          readable
        + use standardized shortcats for bugzillas, e.g. bnc# for
          bugzilla.novell.com
        + remove "." from the end of the line [***]


[*] I do not have any string opinion whether lower or uppercase letter
    is better. Maybe, we should use upper case because there are
    sometimes longer sentences, see
    http://www.grammar-quizzes.com/punc-lists.html

    Also there is problem with shortcats, for example, the script
    sometimes replaces "XML" with "xML" which looks ugly
    

[**] It is not perfect. Developers are inventive and some styles are
     still not handled by the script.

     Note that it is useful also because we add bugzilla numbers from
     the whole commit message; It would look ugly if there are bugzilla
     numbers on two locations, with two styles, in the same sentence.


[***] I think that this it not implemented now.


> * could this behavior be removed? If yes, I can push the patch to master.

I am open for any improvement or changes. I am just curious, why don't
you like the current behavior? ;-)


> Sorry if this mail sounds offensive, it is not intended, of course :-)

It was not offensive :-) I am happy that you are trying to improve the
script.

BTW: Another problem is that it uses "git log tag1..tag2" to show new
fixes in a new release; It is not ideal because it does not detect
cherry-picked commits. It would be better to use "git cherry". Though,
it is more complicated because "git cherry" is not able to print the
full commit message and the name of the author. You would need to call
"git log" for the given commit id to get this information. If you would
want to look at it, it would be great.


Best Regards,
Petr