Hi,
Am 19.01.23 um 12:46 schrieb Andreas Henriksson:
I'd like to explicitly ask if you still want me to upload any
fontconfig change ASAP or if downgrading severity means you think
it can/should wait until after release?
worked around it in libreoffice for now, so your decision in the end.
As I understand the situation something related to subpixel rendering
has changed between 2.13.x and 2.14.1.
Debian offers debconf choices "Automatic", "Always" and "Never" for
which settings to use in /etc/fonts/conf.d/$SYMLINK which are related
to the problem you're reporting.
Fedora on the other hand seems to use "always/only on KDE" (with no
alternatives) as far as I understand the commit you pointed out.
Does that mean we should:
- offer an additional alternative /or/ patch "Always"?
- make then new/patched alternative the default?
- get rid of debconf alternatives and provide one consistent
configuration for all users?
I'd vote for the latter, the deconf thingy is confusing and leads to
problems like this if stuff assumes behaviour
Unless we do the last option, wouldn't it mean any (test) failures
still happen if the user uses any of the other options? (Which would
suggest this is not actually a bug on the fontconfig side to me.)
Probably. Maybe just do what upstreams default is? (Whatever that is).
Or sync with other distros which so far had no problems (aka the patch I
linked), so it's at least consistent between distros.
Fedora actually (as one can see in the linked RH bug) did an update to
released Fedora versions to add that <test>...</test> and TTBOMK Abira
TAGOH is also fontconfig upstream?
In the mail where you tagged this patch you suggested to edit the
subpixel config file which makes me think you're using the Always
alternative, but as I understand the debconf code Automatic is
the (current) default...
Can't be. The buildds (and also my buildd chroot) does not get any
debconf prompt (noninteractive frontend) so default is what I pasted
- without the <test>...</test>.
If you think you have a clear picture of exactly what should happen
(on the fontconfig side) then I would very much appreciate if you
could either send a salsa MR or just NMU (and I'll make sure to import
the NMU-diff into the fontconfig packaging repo).
Unfortunately not, I just try to fix the FTBFS ;)
Regards,
Rene