[PATCH 0/4] Fix and improve Freetype/Fontconfig support

[Matthias Dahl]

Hallo Andrew...

On 16/01/17 15:37, Andrew Hughes wrote:

> I recalled there was a bug about this... turned out it was yours :P

Yes, indeed. I referenced it also in one of my commits. And besides,
a personal need is always a good driving force to fix some bugs. ;-)

> Oh ok. That's not the impression I'd been given. It makes things a lot
> simpler.

Here, I am running everything without any Infinality patches anywhere,
btw. But I can see why people might have gotten the wrong impression.

As a side-note: I have been using those patches for a few weeks now and
personally I prefer the font rendering over the Oracle JDK even. And I
have yet to run into any problems... which probably means nothing.

> I guess we wait and see if anyone hits problems with an old version that's
> still around. That's likely to happen more if we also backport this to 7,
> which it sounds like we should.

Good idea. And backporting to 7 should be as easy as applying the
patches there since in that area, probably nothing has changed - without
really having looked now.

> Well, every setup should really have fontconfig, even without these changes.
> What we need is an option to turn on this enhancement, such as --enable-x.
> What would be an appropriate description for it? --enable-improved-font-rendering?
> 
> The --with-x syntax is meant to take an argument, so --enable/--disable
> seems more appropriate here.

Now I have written several answers for this, each time deleting it again
and rethinking my opinion. That's really a tough one -- and I tend to
have a difficult time naming things, since I always want to be precise
and expressive -- or to put it differently, I am perfectionist in that
regard. ;-)

I was about to suggest "fontconfig-support" with a description along the
lines of "Use fontconfig for font matching and improved FreeType
rendering parameters".

But yours is probably better from a "marketing" point of view and gets
the point across clearer.

One can argue for both, with good reasons. So whatever you decide, I am
fine with it either way. :)

> No, just someone to fix any bugs that come up and generally avoid it
> bit-rotting.

Sure. If anything comes up, assigning the bug to me or put me on cc or
whatever, and I will take care of it.

> Yeah, I'm thinking put it in now, but enable by default for 3.4.0 in April.

Sounds reasonable. I suggest, though, to make it clear in the release
notes / announcement, that this can be used with vanilla Fontconfig and
FreeType, so people actually try it and report back if there are problems.

I will contact the Arch maintainers and see if they want to enable it by
default (or test it) for their packages since they had it enabled
previously. That would give us some feedback. And Gentoo will, as a sure
bet, expose it as an use flag, so I will see what I can do there too for
the default.

> Yes, you'd have to sign an OCA. It's also probably going to have to wait
> until they release 9 and start on 10, as I believe only regression fixes are
> being accepted at the moment.
> 
> Hopefully, a jdk10 set of repos will be forthcoming soon.

Reviewing that whole OCA situation is on my todo and once I have made up
my mind, I will let you know. I am not so keen on signing anything, by
default. So will give that a proper read-through first... ;)

Thanks,
Matthias

-- 
Dipl.-Inf. (FH) Matthias Dahl | Software Engineer | binary-island.eu
 Hire me for contract work: Open Source, Proprietary, Web/Mobile/Desktop