Scarabeus' blag

As summary says in Gentoo you can finally decide which implementation of libav you want, good old ffmpeg named flavor or the new and shiny libav one.

14 months after I introduced it to main tree we finally got it stable almost everywhere (just ppc missing yet as usual :P). We did *censored* of work in order to support both implementations, introducing virtual package, proper (non-conditional) patches everywhere, etc. etc.

So why should you switch your package?

Basically it does not matter for you. We can say it can be done as personal prefference.
We can say that ffmpeg merges from libav tree and throws some experimental patches over it. Thats why my personal prefference is with libav folks as I don’t trust any idea of merging anything. Otoh the ffmpeg is the “orginal” (well most devs moved under libav) upstream thus they should get our trust for what they are doing.

Today from the distro PoV there are some distributions using libav (Debian, Sabayon) and also ffmpeg (openSUSE [external repo called packman]).

As we in Gentoo really are pro-choice just pick yourself and let us know how happy are you with such decision :-P

So how can I switch?

Just edit /etc/portage/package.use* to keep same useflag from media-video/ffmpeg to media-video/libav and run:

emerge -C ffmpeg && emerge -1v libav

After this move you need to recompile all the packages depending on ffmpeg. This can be done with little help of revdep_rebuild or by using portage-2.2 and its sets by running:

emerge @preserved-rebuild

packages that hard require media-video/ffmpeg

As reported in the first comment the mplayer1 does now strictly require ffmpeg. So I did a little check to see which packages hard-require the other implementation and here is the list:

games-arcade/performous-0.6.1 (media-video/ffmpeg)
media-libs/ffmpegsource-2.16.2.1_pre587 (>=media-video/ffmpeg-0.9)
media-libs/mediastreamer-2.3.0-r1 (video ? media-video/ffmpeg)
media-plugins/audacious-plugins-3.0.3 (ffmpeg ? >=media-video/ffmpeg-0.7.3)
media-plugins/audacious-plugins-3.1 (ffmpeg ? >=media-video/ffmpeg-0.7.3)
media-plugins/audacious-plugins-3.1.1 (ffmpeg ? >=media-video/ffmpeg-0.7.3)
media-plugins/audacious-plugins-3.2 (ffmpeg ? >=media-video/ffmpeg-0.7.3)
media-plugins/audacious-plugins-3.2.1 (ffmpeg ? >=media-video/ffmpeg-0.7.3)
media-plugins/audacious-plugins-3.2.2-r1 (ffmpeg ? >=media-video/ffmpeg-0.7.3)
media-plugins/mediastreamer-x264-1.1.7 (media-video/ffmpeg)
media-plugins/mediastreamer-x264-1.3.3 (media-video/ffmpeg)
media-video/dv2sub-0.3 (kino ? media-video/ffmpeg)
media-video/mplayer-1.0_rc4_p20120213 (>=media-video/ffmpeg-0.10)
media-video/mplayer-1.0_rc4_p20120405 (>=media-video/ffmpeg-0.10.2)
media-video/mplayer-9999 (>media-video/ffmpeg-0.10.2)
media-video/transcode-1.1.5-r2 (postproc ? media-video/ffmpeg)
net-libs/opal-2.2.11 (>=media-video/ffmpeg-0.4.7)
net-libs/opal-3.6.8 (ffmpeg ? >=media-video/ffmpeg-0.5[encode])
                    (x264 ? >=media-video/ffmpeg-0.4.7)
net-libs/opal-3.6.8-r1 (ffmpeg ? >=media-video/ffmpeg-0.5[encode])
                       (x264 ? >=media-video/ffmpeg-0.4.7)
net-libs/openh323-1.18.0 (>=media-video/ffmpeg-0.4.7)

I bet most of these could work with libav, so if you are user of such package and want to try libav just edit the dependency to be on virtual/ffmpeg instead of the media-video/ffmpeg and see if it works.

As from today we have new app-officeext category that will be holding various office extensions. You all for sure know that when I say office I mean libreoffice/openoffice extensions. This means that we can now easily add new and shiny ones if you desire so. Please open bugs for your favorite extensions and we will ship them ASAP (also prefferably open stable requests for the current ones so I don’t have to remember to do it :P).

You might be able to find some fancy extensions there already that might help you improve your workflow with libreoffice.

Here is the list I will do this afternoon as they are part of the libreoffice core we never bothered to provide yet:
google-docs ; barcode ; diagram ; hunart ; numbertext ; oooblogger ; typo ; validator ; watch-window ; ct2n (requres two patches from lo tree -> repack).

Apache Office 3.4 released

From other news there is now openoffice 3.4 release made by folks from Apache Foundation. Gentoo will not package this release as it is basically equal to what we would call libreoffice-3.4.1 with some missing features and lacking all the build fixes that were put in lo-3.4.

But by the Gentoo here I just mean that I am epically lazy to solve again all the build issues 3.3 had. So if anyone is willing to maintain and package it we will gladly add it to the main tree (with the named person as maitnainer). You don’t even have to be a developer, there is a nice proxy-maintainer project so poke me mail if you are interested.

What we could also do is to add hardmasked binary version.That stuff contains security issues and is basically the same like old libreoffice/openoffice binaries which tried to contain everything from python to icu. Also for this someone interested is required. So mail me/comment here.

Dictionaries

There will be progress when I find some more time. New simpler eclass for myspell (actually hunspell) dictionaries is in place and I just need to check and verify all the releases. For this I got nice script from Caolan McNamara (LO guy working for Redhat) that contains upstreams and versioning schemes for most of the languages.

This will also be used with utilizing the euscan tool writen by Corentin Chary to watch for newer versions. Where I will also try to bump those with compatible licensing directly in dictionaries repo of libreoffice so you will be able to enjoy the spellchecking even on fancy platforms that none sane would use (riiiight? :-P).

cups dependency

Lots of you lads whined^Wcomplained about the fact cups is actual hard dependency now in libreoffice. With help of one brave lad (Dave Flogeras) we managed to bake configure switch for cups again. It is currently under review on libreoffice mailinglist (not really it already passed and I was just lazy to commit it in which I will do this afternoon). After it is included our live ebuild will again have switch to disable printing completely.

Internal lo extensions

Libreoffice itself has some internal extensions that are built as part of the build process. In history those were never built, or always built, depending on the complexity of their deps.

This now changes. There is new LIBREOFICE_EXTENSIONS use expand variable that can be used same way like VIDEO_CARDS in make.conf.

All of the extensions are already in except reportbuilder, because I was unable to find some java packages (I mailed java team for help here so those hopefully will be identified soon).

If you are interested in contributing to libreoffice consult various comments I put in the live ebuild as I am too sexy^Wlazy to open easy hacks on freedesktop bugzilla. Any help wrt unbundling stuff is welcome. Currently there is even problem with building javascript engine interface because bundled rhino uses too patched debug interfaces (one can either remove it or make work on the currently provided one) so if you like javascript and java you can have fun.

If you speak some fancy language other than english, then this post is exactly for you. While I was working on updating some of the lingua packages in the cvs based on euscan reports I checked their websites to ensure the euscan reports to be correct (in some cases it might miss a bump, or report some beta files). And I hit huge problem, all of those damn pages are writen in native language (eg hungarian spell is in hungarian, dutch in dutch…) which effectively means I can’t read s**t. It might be briliant to get more contributors, but most packagers for the software are developers which just try to make the users happy and do not speak your language at all (google translate is not much help in some cases).

To the point. If you can could you look up dictionaries for your language and check for new versions (ispell/aspell/myspell/…)? If there are some just drop me mail, open new bug and CC me on it, or whatever suits your plan.

Another problem with our app-dicts team is that it is not existing. Currently it is just pva (doing his best) and nobody else did any relevant work on dictionaries since 2k8. We have old support code for openoffice-1/2 and so on. Luckily it does not break anything much, but still it is quite cruft. So if you ever considered being Gentoo contributor this might be easy area how to get involved.

And for the end we need to rework all the myspell-* language packages as they still use old openoffice infrastructure. This was shot down when oracle given all the stuff to apache foundation so all the links are dead. Job here is to search up dict/thes/whatever and set up proper homepage and download locations, maybe asking on irc channels of both projects ( libreoffice and openoffice) to see where they moved it up.

So anyone up for any of the tasks (no, I don’t want to do it because I am lazy, but I have quite lot of stuff within the libreo stuff itself)?

Time has come to admit that 3.4 was not a great release as whole while the 3.5 series provies much much better experience wrt building and having runtime features working.

Continue reading →

Lately people complain that there is no cups useflag on libreoffice.

This in theory changed on 3.5 series where the useflag is present but only as pseudo-optional for runtime because build still requires cups.h. On runtime side it is always optional as it tries to load the lib and just disables the feature if it is not working.

I thought I would ask you guys if anyone is willing to properly ifdef the headers/code in the libreoffice so it is not required for the build either. The fix should be quite trivial as only /core/vcl/unx/generic/printer/ is referencing it.

Anyone up for this? I myself actually have printer and use the cups so I don’t have ANY motivation to do this but I can review/commit what you guys create.

openoffice

Seems like Apache office is nearing to some 3.4 release. Would anyone be interested in working on that? It contain most of the build system bugs that were fixed on libreoffice so I don’t want to touch it even with a long stick, but maybe somebody would be interested enough to contribute it in Gentoo :)

As you probably know I am using Gentoo hardened on stable flavor. Lately it led me to kind of thinking that in the main tree we have lots and lots of packages in testing only and people spam their /etc/portage/package.keywords* with various packages they need for their stable life.

I think we can somehow improve and YOU as users should do something about it by filling stabilisation requests. Mostly it is because developers only care about testing so bugs are fixed there, but nobody ever bothers to open the stabilisation bug. Resutls vary but one of visible ones is that we sometimes have packages that have last stabilisation in 2k9 even tho in testing there is last update this year.

This is often because bug gets only fixed by non-maintainer, qa, or just by somebody going around and fixing the issue for himself. This can lead to fixes landing to testing tree, but from then on nobody cares to propagate them properly for us puny folks in stable.

Easy way how to work around this is simple opening stable requests on packages you think you want. If maintainer is not responsive in 30 days (aka saying anything there) you as users can CC arch teams yourself (i suppose even one arch team only if you are unsure which ones is enough because they will cc the rest) and get the package into stable tree.

The stabilisation of non-maintainer packages often means more work for the AT/devs as you really should check the bugs and see if there are no crashes/patches around and try to apply them, I know you probably want to say that you don’t want to do anything more than changing the keywords but it improves the overall status :)

On the testing front I find it boring to search for user patches on bugzilla. I think we should do something like mailinglist where anyone can sent the patches and any developer could include those patches to main tree if he is using the package. This could speed up user inclusions a lot and also if something important for someone is not included once, he can sent the mail again.

Libreoffice

The 3.5.1 is closing by to arrive tomorow for you guys. This version should be for general consumption so lets see if we can stabilise it as I would really like to get rid of nsplugin and password bugs that are present in 3.4.5. Also for 3.4 series there is 3.4.6 en-route in next 14 days, but it won’t bring up much fixes.

First of all let me just say huge thanks for everyone who donated.

Here is list of people which I am aware of their identity:
David Abbott
Chris Fox
Olaf Freyer
Barrie Backhurst
N Wright

UPDATE
One more person contributed after I wrote the post:
Patrick Nagel

Together you guys managed to donate me 75% money for the disk and as I expected not to get above 50% I even bought the better one (Vertex instead of the Agility).
So you all get few free bugfixes when you mail me with your breakages. This requires me knowing the stack you want fixed of course, I can’t fix Gnome.

Anyway for rest of the readers there is no more any need for donations (if you want you can tho :P).

So what did I set up for your money worth?

Linux arcarius 3.2.2-hardened-r1 #2 SMP Tue Feb 14 18:29:28 CET 2012 x86_64 Intel(R) Core(TM) i7-2600K CPU @ 3.40GHz GenuineIntel GNU/Linux

If you are running on non-hardened machine you should really consider migrating to hardened, it works pretty well. Also really consider running stable instead of testing on your machines and opening stabilisation requests where needed. For me current keywords are this:

scarabeus@arcarius: ~ $ cat /etc/portage/package.keywords/*
=sys-devel/gcc-4.6* **
<sys-boot/grub-9999 **
media-video/libav
media-libs/x264
app-office/libreoffice
app-office/libreoffice-l10n
=dev-cpp/libcmis-0.1.0 ~amd64
=sys-apps/portage-2.2* **
=media-video/vlc-2.0*
dev-db/virtuoso-odbc
dev-db/virtuoso-server
sys-libs/zlib

I need the gcc for i7 and testing of libreoffice. The lo obviously is needed testing for development. Portage for @revdep-rebuild (can’t imagine life without it anymore) and vlc-2.0 where I am translator so I need it for the translation testing. Rest are just dependencies for those, where I plan to open stablereqs asap.

What about the disk layout?

scarabeus@arcarius: ~ $ LC_ALL=C df
Filesystem                       Size  Used Avail Use% Mounted on
rootfs                            89G   12G   78G  13% /
udev                              10M  284K  9.8M   3% /dev
/dev/sda4                         89G   12G   78G  13% /
rc-svcdir                        1.0M  112K  912K  11% /lib64/rc/init.d
cgroup_root                       10M     0   10M   0% /sys/fs/cgroup
tmpfs                            7.9G  892K  7.9G   1% /dev/shm
/dev/sda1                        9.6G  4.9G  4.2G  55% /var
/dev/sda3                        3.9G  210M  3.5G   6% /var/log
/dev/sda2                        9.6G  4.2G  5.0G  46% /usr/portage
tmpfs                            2.0G   24K  2.0G   1% /tmp
tmpfs                             10G  278M  9.8G   3% /var/tmp
192.168.137.10:/home/data/       2.7T  1.5T  1.3T  55% /home/data
192.168.137.10:/home/scarabeus/  2.7T  1.5T  1.3T  55% /home/scarabeus/remote

I put KDE there and right away fixed few lo-related bugs so we got from 52 to 45 bugs and to 3.5 final release (propagates itself as rc3). So now again back to clearing live ebuild and preparing new stabilisation list. You can all stalk my stats here

What I for sure won’t do: reassing all the bugs from m-n back to office as it is just bugzilla spam.

Anyway it took me 5 hours from 0 to this state. We can say that Gentoo install is pretty straight forward.
The only problem I had was my nfs setup, which was even on server side, thanks to prometheanfire for the solution.

Sitrep

I’ve been quiet lately due to the stream of changes around me. So shortly I am no longer only user of my desktop nor the major user of it. Thanks to that and issues that Gentoo showed while used by others I simply was unable to afford doing 24/7 tech support for issues with it main disk was sweeped with windows 7 and gentoo was moved to old 250 GB hdd that was long due after its survival date. This disk sadly died, so my remaining Gentoo install is HTPC which has no power to actually build libreoffice in any timely manner (what can one expect from Zacate platform).

Given to that I retired from office team in Gentoo and expected others to enjoy the building the beast. Which should not be that hard as I still work on it as upstream, I just don’t have the Gentoo part. As you can see the mail I was quite wrong as there is nobody actually willing to pick this thing up.

So I need new HDD and given to nowday prices for this components I really don’t want to buy it myself, so anyone wants to donate me some small ssd (30GB or larger as gentoo install + debug symbols take around 20 here [ggdb is huge])? On the other hand as council members behave really warmly these days (I would not care if normal dev said something like this but Markos is one of 7 elected ones picked by everyone) I think I should just ignore Gentoo and go ahead look around for some nicer distro…

Now back to interesting stuff.

libreoffice 3.5 release

3.5 will contain the lot of shiny stuff. Current 3.5.0.3 has fixed nsplugin and various other tiny issues like lpsolver and so on.
It still contains lot of tiny bugs everyone can see here and there, but they are getting fixed as we go so 3.5.1 or 3.5.2 will be ready for corporate environments again.

libreoffice trunk

There is steady stream of patches going into master branch at git.fdo and one of most notable thing is that we no longer use set_soenv perl script and most of the things are converted to gbuild (less paralel make issues, more compilance to MAKEOPTS). Other than that everything is same shiny like in 3.5 so no big difference on usability yet.

VLC translations

In mid-january it was brought to my attention that I am still listed as the one responsible for czech translation at VLC despite translating few strings in 2009. So I started complete overhaul using the KDE translation memory to improve it and remove all the wrong translations (you would not believe how much wrongly translated strings i found so far [I am at 50% of reviewing the current state]). So hopefully I will get this mostly done for 2.0 release so even Czech people can enjoy nice and translated media player.

The current status:

scarabeus@bugaboo: ~/l10n $ LC_ALL=C msgfmt --statistics vlc_new_cs.po
1766 translated messages, 4120 untranslated messages.
scarabeus@bugaboo: ~/l10n $ LC_ALL=C msgfmt --statistics vlc_cs.po
2140 translated messages, 1942 fuzzy translations, 1804 untranslated messages.

Gaming and other stuff

Anyone wants to play Battlefiled 3 or Counterstrike (actually anything you seen on the screenshot above :P)?

KDE release party in Prague

There will be release party for the above stuff in Prague on 24.2. at SUSE Linux offices and I will be giving some talk about KDE and LO (at least Tampakrap convinced me I should) so if you want to meet me (and other more cool KDE hackers) you should show up.

Update

As few people contacted me already I think i would like to buy one of these (one, two) or their smaller variants. If you want to realculate the price for Euros/USD divide by 25/17 respectively.

So if you want to donate feel free to sent money on paypal or mail me if you would rather want IBAN.
I will post results as another blog at point when I have enough money or in 30 days whatever will be sooner.

Lets start with something slightly different

Tomorow there will be 3.4.4 release.

• It brings us less bundled stuff on gentoo side thanks to ebuild cleanup (noting noteworthy upstream on that matter).
• Fixed qt4 build on qt-4.7 (YAY see my previous blog) and fixed build with 4.8.0
• Lots of internal fixes I am lazy to mention

This is the last 3.4 release with upstream love. There also will be 3.4.5 after 3.5.0 release, but it is for 2012, nothing shiny will happen in this year for that branch.

Now for all awesome candies in master

I worked on nsplugin building with system libraries (xulrunner/firefox/npapi-sdk) which worked great, so now it is enabled by default and built against the system headers (note that the nsplugin was build by default even before and the useflag was quite bogus in gentoo).

The mozilla switch that many of you can see there is used for mozab and ldap backend stuff. It sadly requires INTERNAL mozilla to actually work, so I disabled it in Gentoo for now (both moz stuff and ldap) as it does not work. I am being quite skeptic about fixing it for 3.5.0 as freeze is comming around 5th Dec if my memory serves well.

The only issue I still didn’t fix is that SampleICC bundling which is quite hard to do due to patches/magic done there in DMAKE…

Yet again completely unrelated ending section

Seems like I really enjoy to jump from one topic to another…
Anyway I did add really cool game to main tree which I think you should try out.
The package is games-strategy/openxcom which is opensource implementation of UFO: Enemy Unknown (first X-Com based game). It is still alpha quality as some research is not possible, manufactures are being done, ufopedia is not working, but you already can try out shooting ufos and doing missions.
The slightly problematic part is that the game does not provide the data files so you need the orginal game. Other than that it is pretty awesome it works on linux natively instead of requiring dosbox. (So far coolest stuff is that cheating is really simple by editing rulesets in the saved game :P [pistol that decimates whole building is pretty cool :P])

This week at work I got quite frustrated for libreoffice crashing so often for me when I was using them in KDE.
So I decided to do what any lazy slacker would, I went to Lubos Lunak and showed him nice and easy reproduce case.

Thanks to that Lubos wrote simple check that made the problem go away and I did backport the thingie to 3.4 branch.
So with 3.4.4 release this will be included in Gentoo (already is availible under 3.4.9999).

I am just sharing this to show that we don’t just work on new features but we really try hard to fix all those usability issues, but it ain’t always easy as we have bazilions of open bugs going back to really old times :)