Arch Linux

The poppler-qt package has been renamed to poppler-qt4. I've updating this package, but if it causes any issues for anyone, simply remove this package, allow the standard poppler-qt4 package to be installed, and then install this package again to replace it.

poppler 0.26.0 is now in extra

0.24.2-2 brings a fix to an issue where certain documents would be rendered with pages stretched and/or cropped.

0.24.1-2 brings a fix to an issue when zooming in beyond 300% in Okular.

https://github.com/giddie/poppler-qt4-cairo-backend/issues/1

My bad; 0.22.5 is a very simple bump. It's 0.23.2 that required more work, but that's done now, so when Poppler 0.23.2 hits the Arch repositories, I'm now ready :)

Boo! It looks like the update to poppler-0.22.5 will not be a simple rebase. Some rework of the patchset is required. It could take me some time to get around to this, I'm afraid.

@slaha => This happened because the "poppler" and "poppler-glib" packages were updated on your system. You might want to add them to IgnorePkg in /etc/pacman.conf to avoid accidentally updating them before I have a chance to update this package.

Hi, there is an issue after update. (I don't exactly know what update causes that, but I do update to KDE 4.9.5 and poppler to 0.22.0-1). After upgrade okular cannot open any PDF anymore. As solution I installed poppler-qt from repository.

It turns out it was only a small API tweak that was causing the breakage. Everything seems in order now.

Sorry, with this being a new feature release, the patched source isn't compiling nicely. I'll need to look at this. Hopefully I'll get a working PKGBUILD out in the next few days.

Thanks; it's great to hear from someone who's pleased with this :) With the latest changes to selection in Okular in 4.8, it looks like the only major outstanding issue has gone away. In previous versions, selection was rather broken when using the Cairo backend in Poppler, but so far I haven't noticed any of the old graphical corruption that was happening before.

Depending on how outspoken you're feeling today, you might consider letting the developers know you approve of this approach to subpixel rendering by commenting on the Bugzilla report: https://bugs.freedesktop.org/show_bug.cgi?id=25240

Finally with this (Okular + Infinality + poppler-qt-lcd) I have a working subpixel capable PDF reading solution.
Thank you a lot. This should be the default!

These patches don't work for me :(
https://bbs.archlinux.org/viewtopic.php?id=121227

It'll take me a few days to make some necessary changes to the patches, as the Poppler Qt4 wrapper API has changed slightly in 0.16, and I need to check everything's still OK.

For now, I'm removing the patch that activates the slight hinting, as it's doing more harm that good. It definitely seems as though Cairo is doing something weird here; maybe ignoring fontconfig? I'm using the respect-fontconfig patch for Cairo, but I think Poppler uses some unusual Cairo setup internally.

You're right, it is worser, especialy with PDF at Ocular seems to me. Hopefully, we'll overcome.

Right, this should build OK again now. The main issue was that Cairo 1.10.0 has removed the ability to manually select the subpixel filter, but I think FIR5 is the default anyway. I have noticed that the "slight" hinting seems to be a bit more aggressive now. I'm not sure why, and I'm not sure I like it. I've left it enabled for now, though.

Sadly, it seems an increasing number of documents are not properly subpixel-rendered in the Cairo backend, even with these patches. I'm afraid that issue is for other people with more understanding of Cairo internals to solve. (See here: https://bugs.freedesktop.org/show_bug.cgi?id=3307.)

OK, I've just got my system back to pretty much the same standard it was at before. (See https://bbs.archlinux.org/viewtopic.php?id=105839.) Hopefully over the next few days I'll have time to try fixing the poppler patches.

There was some conversation about LCD filtering at aur-general mailing list, a guy states, that he has LCD filtering without using -lcd packages, you may want to ask him for more details.

I've also replaced cairo-lcd with the new cairo and am madly trying to find a way to get fonts looking nice again in cairo-based apps :(

Thanx for a quick response. Sadly when hunting seemingly related insonsolata font problem, I upgraded the cairo lib and uninstalled -lcd patched packages, but I'll manage without it for now.

Thank you for bringing this to my attention. Unfortunately, it appears that the patches I´m using against the Cairo backend don´t yet work with Cairo 1.10.0. I´ll have to look into this. In the meantime, I suggest sticking with Cairo 1.8, if possible.

Hello,
you should remove cairo-lcd dependency in favor of cairo, as it is said [https://bbs.archlinux.org/viewtopic.php?id=105591] that cairo 1.10 now includes lcd patch.

I'll stay with version 0.12.4-1 until things are settled.
There was never any problem with that version.

For the record, this is the bug that we need to follow for progress on subpixel rendering in the Cairo backend:
https://bugs.freedesktop.org/show_bug.cgi?id=3307

All the people there will be working with the glib wrapper and Evince, but if they can get subpixel rendering working, it'll work for us too :)

I've had a look at the document, and for me it renders OK. The main body text is not subpixel-rendered, the text in the boxes is rendered correctly, and I see no artifacts like you're getting. I wonder if something happened to your cairo-lcd? You could try recompiling it to see if it makes a difference.

OK here is the link (Just click »Download (4671K)«):
http://docs.google.com/fileview?id=0B2Ny85tHSOg_OTllMmM5NDItNzZmOS00ODg1LWEwNjItN2ZkMTFlYTgyMmQz

And BTW the same document displayed fine (subpixel rendering) with poppler-qt-lcd 0.12.4-1.

That's odd; I'm using cairo-lcd from AUR, but I don't see those artifacts. Are you able to give me a link to that document to see if I can reproduce the issue? I do have several documents for which subpixel rendering isn't working though. Unfortunately, this will be a problem with the Cairo backend itself. There may be a patch on Bugzilla at Freedesktop.org; I'll have a look when I have some time.

I do wish it weren't necessary to hardcode the subpixel order, but I'm hoping that someone somewhere will add proper subpixel rendering support to the Cairo backend before long, as I know very little about Cairo myself.

Thanks for your quick reply.

I've tested your new PKGBUILD and it does not fully work here unfortunately.
Here are two screenshots highlighting the problem:
http://omploader.org/vNHE1Yg
http://omploader.org/vNHE1Yw
The first one shows an pdf using aur/cairo-lcd. As you can see only the grey
box is rendered correctly the other text is corrupted.
The second screenshot shows the same pdf with aur/cairo-cleartype. This is
better, but only the text in the grey box is rendered using subpixels, the rest
of the test is rendered with plain greyscale aliasing.

(BTW I use my monitor in portrait mode, so I've hacked your
0003-Forcing-subpixel-rendering-in-Cairo-backend.patch to use
CAIRO_SUBPIXEL_ORDER_VRGB)

Thanks for notifying me that 0.14 is in testing. I've updated the patches on Github, and prepared a new PKGBUILD, but I'd rather not upload it here until 0.14 is in extra. In the meantime, you can get the updated PKGBUILD from http://giddie.homeip.net/poppler-qt-lcd-0.14.0-1.src.tar.gz. Please let me know if it works for you, so I can be confident it's OK for the masses when 0.14 hits extra :)

Since poppler 0.14.0-1 is in testing, it would be nice if you could update this package.