git: a9fc14fbf445 - main - newvers.sh: add support for gitup(1)

Mateusz Guzik mjguzik at gmail.com
Thu Jan 21 15:16:38 UTC 2021


On 1/21/21, Jessica Clarke <jrtc27 at freebsd.org> wrote:
> On 21 Jan 2021, at 14:32, Mateusz Guzik <mjguzik at gmail.com> wrote:
>> On 1/21/21, Ed Maste <emaste at freebsd.org> wrote:
>>> On Thu, 21 Jan 2021 at 05:37, Mateusz Guzik <mjguzik at gmail.com> wrote:
>>>>
>>>> I'm getting an error with this:
>>>>
>>>> $ make -s buildkernel MODULES_OVERRIDE=""
>>>>
>>>> /tank/users/mjg/src/freebsd/sys/conf/newvers.sh: git: not found
>>>>
>>>>> +     if [ "$(git rev-parse --is-shallow-repository)" = false ] ; then
>>>                  ^^^
>>> Looks like this bare git should be $git_cmd, can you confirm that
>>> change fixes it?
>>
>> This fixes the problem at hand, but a new regression is shown:
>>
>> $ strings kernel | grep 'FreeBSD 13.0-ALPHA1'
>> FreeBSD 13.0-ALPHA1 #47 c256181-gb06e91d1b700: Thu Jan 21 14:31:35 UTC
>> 2021
>>
>> branch name is no longer present.
>
> See the commit message, though I'd argue it's not redundant in a lot of
> local development cases (i.e. anything other than main, stable/*, etc),
> and is actually rather useful.
>

right


    While here, drop the redundant branch name from the git output and don't
    count commits in shallow clones.

That's definitely a regression. If the branch name is really seen a
spurious (why?), then perhaps it can be only skipped if name == main

-- 
Mateusz Guzik <mjguzik gmail.com>


More information about the dev-commits-src-all mailing list