hash_delete safe to remove?

hash_delete safe to remove?

[Collin Funk]

A month or two ago I updated gnulib in GNU patch to fix some build
failures. I also noticed some deprecation warnings from 'hash_delete'
and changed them to 'hash_remove'.

I forgot to ask here, but should this be removed now? Or bump the
date? :)

In lib/hash.c:

/* Same as hash_remove.  This interface is deprecated.
   FIXME: Remove in 2022.  */

Collin

Re: hash_delete safe to remove?

[Bruno Haible]

Collin Funk wrote:
> A month or two ago I updated gnulib in GNU patch to fix some build
> failures. I also noticed some deprecation warnings from 'hash_delete'
> and changed them to 'hash_remove'.
> 
> I forgot to ask here, but should this be removed now?

Has there been an official 'patch' release that includes the change,
already? If not, it's too early: Some people (or distros) like to
take the git checkout of the latest release and rebuild the generated
files with the newest gnulib. There will be some things to update, sure
(like the 'execute' argument list). But the fewer problems there are,
the better. Even if we don't promise that there are no problems.

In other words: Each backward-incompatible change (-> mentioned in the
NEWS file) can trigger problems. Let's try to minimize the trouble.

Bruno

Re: hash_delete safe to remove?

[Collin Funk]

On 4/13/24 5:50 PM, Bruno Haible wrote:
> Has there been an official 'patch' release that includes the change,
> already? If not, it's too early: Some people (or distros) like to
> take the git checkout of the latest release and rebuild the generated
> files with the newest gnulib. There will be some things to update, sure
> (like the 'execute' argument list). But the fewer problems there are,
> the better. Even if we don't promise that there are no problems.

That makes sense. Thanks for the explanation. No there hasn't been a
release yet. My original reason for updating Gnulib there was build
errors that were fixed in a subsequent commit [1]. That also involved
updating the argument list to 'execute'.

> In other words: Each backward-incompatible change (-> mentioned in the
> NEWS file) can trigger problems. Let's try to minimize the trouble.

Sounds good.

[1] https://lists.gnu.org/archive/html/bug-patch/2024-02/msg00000.html

Collin