On Sun, Jan 17, 2021 at 5:46 PM Victor Yegorov <vyegorov@gmail.com> wrote:
>
> вс, 17 янв. 2021 г. в 17:19, Magnus Hagander <magnus@hagander.net>:
>>
>> > First thing I've noted:
>> >
>> > https://git.postgresql.org/cgit/postgresql.git/commit/960869da0803427d14335bba24393f414b476e2c
>> >
>> > silently shows another commit.
>>
>> Where did you get that URL from?
>
>
> I've made it up manually, comparing cgit and gitweb links.
>
>
>>
>> And AFAICT, and URL like that in cgit shows the latest commit in the
>> repo, for the path that you entered (which in this case is the hash
>> put int he wrong place).
>
>
> Yes, that's what I've noted too.
>
>
>> I guess we could capture a specific "looks like a hash" and redirect
>> that, assuming we would never ever have anything in a path or filename
>> in any of our repositories that looks like a hash. That seems like
>> maybe it's a bit of a broad assumption?
>
>
> I thought maybe it's possible to rewrite requests in a form:
>
> /cgit/*/commit/*
>
> into
>
> /cgit/*/commit/?id=&
That would break any repository that has a directory called "commit"
in it, wouldn't it?
That said we might be able to pick it up as a top level entry only,
because those subdirs would be expected to be under /tree/*/commit/*.
But we could also not do /cgit/<one level>/commit/* -- for example
https://git.postgresql.org/cgit/postgresql.git/commit/src/backend/tcop/postgres.c?id=960869da0803427d14335bba24393f414b476e2c
is a perfectly valid url to show the part of the patch that affects
just this one part of the path.
--
Magnus Hagander
Me: https://www.hagander.net/
Work: https://www.redpill-linpro.com/
