On 11/3/24 18:39, Junio C Hamano wrote:
> Ignacio Encinas Rubio <ignacio@xxxxxxxxxxxx> writes:
>
>> I can't think of any room for improvement other than integrating
>> hostname (or a custom hostname) into git and using it in the tests, but
>> I doubt it is worth it.
>
> Ah, that is a thought. We have t/helper that builds "test-tool"
> just for that, and exposing the output of xhostname() does sounds
> like a reasonable way to go. It would roughly involve
Great! I hadn't noticed "test-tool". Just to double-check, what name do
we want to use for this? xhostname, hostname, xgethostname, gethostname?
If I didn't miss something, the only place the test use hostname is in
$ git grep '$(hostname' t/
t/t6500-gc.sh: hostname=$(hostname || echo unknown) &&
as you previously pointed out. So my plan is:
1. Extend test-tool, migrate t6500-gc.sh to test-tool xhostname(*)
2. Update my v2 to use "test-tool xhostname(*)"
(*) or however we want to name it
> * Add t/helper/test-xhostname.c that defines cmd__xhostname() and
> writes the result of calling xhostname() to its standard output.
>
> * Plumb it through by adding it to a few places:
>
> - t/helper/test-tool.h wants the extern definition.
> - t/helper/test-tool.c wants it in its cmds[] array.
> - Makefile wants to list it in TEST_BUILTIN_OBJS
>
> * Then use "test-tool xhostname" in your tests, instead of
> "hostname".
>
> You can run
>
> $ git grep chmtime ':!t/*.sh"
>
> to find places that needed to be touched when a similar internal
> tool "chmtime" was added.
Thank you very much for the pointers, they were very helpful!
