Reverting the mold linker

[Volker Krause]

Come on everyone, we can do better than this. This was entirely preventable:

Changes like this should be done by an MR (at least one for a meta-review, not 
necessarily one per repo). That goes both for the orginal change and the 
revert.

Even more importantly: *Talk to each other*. There were concerns discussed 
about this on Matrix shortly after the initial change 6 weeks ago, but that 
never made it to Laurent, so unsurprisingly nothing happened.

I also failed on this point, when I got linker errors due to this I 
uninstalled mold as a quick fix and didn't follow up later when I had the time.

We probably all agree on the motivation on the original change, reducing build 
times. How and where to best implement this is a discussion we should have had 
6 weeks ago.

Regards,
Volker

On Tuesday, 6 October 2026 06:43:00 Central European Summer Time Laurent 
Montel wrote:
> Le 06.10.2026 00:43, Albert Astals Cid a écrit :
> >> But it seems that you don't care to apply your rules to you.
> >> 
> >> 
> >> I didn't know that you maintain something in pim*
> > 
> > Please, let's try to keep this discussion civil.
> 
> be civil will be first to discussing with maintainer no ?
> 
> > Being a maintainer doesn't necessarily mean that a particular decision
> > is the
> > right one for the project.
> 
> so a discussion without main developer is ok for you.
> 
> 
> 
> As I can see he knows the email about mailing list (kde-pim), so if he
> wanted
> to discuss or inform about it he could use this ML no  (but not 10
> seconds before
> reverting) ?
> 
> 
> So there is two rules:
> - One rule from him in framework: "We must use MR, otherwise I reverse
> without discussion."
> - One rule for him in other project: "I don't care about maintainer",
> even if I don't participate
> really to the project.
> 
> > From reading both the comments here and in Matrix it seems that forcing
> > mold
> > in cmake is causing problems for some people.
> 
> Where is bugs report about it ?
> Where is the mail in ML about it?
> 
> 
> I didn't see CI broken about it. I build from year with mold without
> problem.
> 
> There was an option for disabling it "-DUSE_MOLD_LINKER=OFF"
> 
> It will use only if system has it. It's not a hard dependancy.
> 
> > For those who want to use mold, setting the linker should be as simple
> > as
> > adding -DCMAKE_LINKER_TYPE=MOLD to their scripts.
> > 
> > From my perspective, removing the forced use of mold seems like a
> > reasonable
> > approach: it addresses the problems people are experiencing while still
> > allowing those who want mold to enable it relatively easily.
> 
> This revert was not ok as he didn't revert the change in CMakePreset =>
> now
> we can't use mold easily.
> 
> 
> Regards
> 
> > Best Regards,
> > 
> >   Albert


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 232 bytes
Desc: This is a digitally signed message part.
URL: <http://mail.kde.org/pipermail/kde-pim/attachments/20261006/b967acbe/attachment.sig>