Re: [GSOC PATCH] t7003: modernize path existence checks using test helpers

From: SoutrikDas <valusoutrik@gmail.com>
Date: Feb 10, 2026, 18:14 UTC
Message-ID: <20260210181445.49380-1-valusoutrik@gmail.com>
In-Reply-To: <xmqqpl6d4wjh.fsf@gitster.g>

Show 6 quoted lines

> All the checks involving "is-missing" are now stricter than the
> original, in that they used to allow "dir" to exist as long as it is
> not a directory, etc., but if we audited the code that leads to
> these tests can never create a "dir" that is a regular file or
> something that is not a directory (which *I* did *NOT*, but
> presumably you have already done so?

At the time of sending the patch v1, I did not do so. Sorry about that. Now I ran the test from start to 11, since test 12 was the one with two of those risky changes, ie `! test -d dir` and `! test -d diroh/dir`

and after doing that I can confirm that there is no non directory dir present before test 12 starts. Neither is there a non directory dir inside `diroh`

This was the output of ls -la

drwxr-xr-x 14 soutrik staff 448 10 Feb 23:16 . drwxr-xr-x 4 soutrik staff 128 10 Feb 23:14 .. drwxr-xr-x@ 14 soutrik staff 448 10 Feb 23:16 .git -rw-r--r-- 1 soutrik staff 2 10 Feb 23:15 A.t -rw-r--r-- 1 soutrik staff 2 10 Feb 23:15 B.t -rw-r--r--@ 1 soutrik staff 128 10 Feb 23:16 backup-refs -rw-r--r--@ 1 soutrik staff 2 10 Feb 23:15 C.t drwxr-xr-x@ 3 soutrik staff 96 10 Feb 23:16 diroh -rw-r--r--@ 1 soutrik staff 2 10 Feb 23:16 doh drwxr-xr-x 4 soutrik staff 128 10 Feb 23:15 drepo drwxr-xr-x@ 6 soutrik staff 192 10 Feb 23:16 drepo-tree -rw-r--r--@ 1 soutrik staff 2 10 Feb 23:15 E.t -rw-r--r-- 1 soutrik staff 2 10 Feb 23:15 G.t -rw-r--r-- 1 soutrik staff 2 10 Feb 23:15 H.t

And this was the ls -la in the `diroh` directory

drwxr-xr-x@ 3 soutrik staff 96 10 Feb 23:16 . drwxr-xr-x 14 soutrik staff 448 10 Feb 23:16 .. -rw-r--r--@ 1 soutrik staff 6 10 Feb 23:16 D.t

One thing is that I am not sure if what I did is ... the correct way to test this kind of thing... I just copy pasted all the test commands into a big .sh file , did a git init -b main in a temp folder and ran the .sh file from there. And then observed the changes. That big .sh file : https://pastebin.com/9QQD7qYA

> ---if so that is worth noting in
> the proposed log message), then "test ! -d dir" that is rewritten to
> "test_path_is_missing dir" is actually a _better_ test.

I am sending the v2 patch after this message. Should I have just put this whole thing into the v2 patch cover mail ?

Best, Soutrik