Re: m4_version_compare and new version numbering

[Jim Meyering]

Eric Blake <address@hidden> wrote:

> According to Jim Meyering on 11/3/2007 5:01 AM:
>>   $ cat --version|head -1              cat (GNU coreutils) 6.9-375-3e3f8
>>   $ cat --version|sed s/-/./|head -1   cat (GNU coreutils) 6.9.375-3e3f8
>>   $ cat --version|sed s/-/./g|head -1  cat (GNU coreutils) 6.9.375.3e3f8
>>
>> I'm inclined to switch coreutils to use the all-'.' version.
>> Any preference/arguments for leaving the final '-' ?
>
> I kind of like the middle approach, with '-' before the SHA1; the version
> is 6.9.375 (ie. 375 changesets since 6.9), while the -3e3f8 offers
> additional information on which commit lands in position 375, but doesn't
> add any additional versioning granularity.
>
>> Anyone know why git adds the 'g' prefix to the SHA1 in git-describe?
>> If there's a good reason, maybe git-version-gen should not remove it,
>> after all.
>
> I'm okay with keeping the 'g' prefix out for brevity, provided you then
> use the separator '-' to keep the distinction of the role of the suffix as
> a non-monotonic string that represents the commit id.

Good points.  Assuming sane tools.
Consider me convinced, unless the git folks have
a good argument to the contrary.  I posted there:

  http://thread.gmane.org/gmane.comp.version-control.git/63255