On Sat, 28 Jun 2025 14:19:01 +0100
void <[email protected]> wrote:

> On Sat, Jun 28, 2025 at 02:00:41AM +0100, Jamie Landeg-Jones wrote:
> 
> >He was asking about flavours. If you're using options, You can actually do
> >it globally. I'm not sure if options will work in "voids" use-case, but
> >this is how it's done:
> 
> The issue seems to be that, for some qt ports, qt6 is an option.
> For others, it's a flavo(u)r.
> 
> I just want everything that is qt, to be qt6. If a port is
> qt5-only, I'd like it to fail (to be built). If it's just qt
> and has neither an option nor a flavo(u)r, (i.e. it just uses
> what qt is there, in other words, and in this case, qt6),
> then it can be built.
> 
> Other ports have default versions that can be set/overridden
> I'm surprised this doesn't exist in the same way for qt 
> and the option/flavo(u)r conundrum.
> 
> For now, I'll set OPTIONS_UNSET+=QT5 and OPTIONS_SET+=QT6
> and set @qt6 for all the ports where it's not an option but
> a flavo(u)r. It's not a showstopper; it's more of an academic
> "why is this not like the others".
> -- 

Yeah, for ports, it would work without issue. If the option is
nonexistent, it would be simply ignored.

But for flavor, you could see a plenty of errors, IIRC, basically just
annoying, though. It appeared at the very beginning of builds and (not
precise but) mostly "false positives", but could cause building the
dependencies of errored-out ports not to be started.

Hope my script (and its README.md shown) linked below help a bit more to
understand.

  https://brew.bsd.cafe/TomAoki/sh_scripts/src/branch/main/poudlist-all

I'm using this script (actually setting more in the BLOCKLIST) to
generate list of all installed ports with flavors, excluding too large
leaf ports to use with -f option of `poudriere bulk`.

If you want to switch qt5 ports to qt6 in existing installation and
building pkgs with poudriere locally, you can open the generated list
in an editor you prefer, search for "@qt5" and replace it with "@qt6"
would be convenient.


-- 
Tomoaki AOKI    <[email protected]>

Reply via email to