Diego Petten
Archives for: May 2005
23 May , 2005
YES! Finally!
Ok here we are, always more and more near to the release of Gentoo/FreeBSD. Today I submitted a patch for portage which makes env-update work fine on FreeBSD with its ldconfig.
I also cleaned up pamd stuff in freebsd-* packages, and created a new profile for x86-fbsd so to be prepared for new arches when they'll be possible.
As you can see, I'm adding screenshots of KDE running on my (headless, for who is wondering why the krdc stuff above) g/fbsd box.
22 May , 2005
Finally binutils works
Ok, thanks to our loved Vapier, finally we have working binutils under Gentoo/FreeBSD. This makes possible to hope for our first stage soon :)
Trying to summarize what's still needed before the real release: the PAM problem is being fixed pace-by-pace, now we have pam eclass and most of the ebuilds which uses pam_stack are already fixed or pending fixes; root:root is still a big issue, but I hope it can be fixed soon, very very soon; metamail is still blocking my hylafax installation, but I hope to get that fixed as soon as I have time; cups has issues for the missing lp user/group, I'll report the bug soon; cp -a ebuilds needs to be fixed soon.
Unfortunately there are still a couple of problems which prevent me from having a full working G/FBSD system with X11-over-VNC. The first of them is that I'm not able to start X as user.. I'll work that out doing a bit of debug on startx command. The second of them is that qt doesn't build as I hoped... it fails because it needs to know the platform you are in... I hope I can fix in a couple of minutes.
Well time to return hacking.... bye!
19 May , 2005
Most likely problems in ebuilds porting
As binutils is still not fixed (vapier provided me a patch that should embody all the changes done by freebsd, but still doesn't work so I think the problem is in a missing script), and so I can't really fix xorg, I'm trying to fix other problems with G/FBSD porting.
There are a couple of major issues with it, which still needs to be fixed, which I want to state here so that other devs can read and know how to deal with them in the future :)
First problem is PAM-related: pam_stack.so module used to mimic other auth configuration files is just an hack done by RedHat that works just on Linux-PAM and not on OpenPAM which is the implementation used by FreeBSD. To do the same thing, OpenPAM and Linux-PAM 0.78 uses the include directive, which works fine on both systems. Unfortunately, too many files in the tree depends on pam_stack, and stable Linux-PAM doesn't support include directive, so they can't just be fixed in the main file, but needs a revbump and a parallel file in the mean time.
I'm taking care of this problem filing bugs aobut packages which uses pam_stack.so.
The second problem is with many ebuilds which resets permissions.. they use root:root as owner:group, but root group doesn't exists on FreeBSD (so neither on G/FBSD), and this needs to be changed to root:whell which works fine on all Gentoo systems.
I've just created a list of ebuilds which suffers from the problem, and they are many more than you can imagine. I hope to be able to fix most of them, but if I start reporting one by one it will take too many bugs so I 'll just open one bug CCing the maintainers while I found them. I'll start tomorrow with that.
The third problem is with "cp -a". cp command accepts -a parameter only on GNU's version, so using it with non-gnu's cp will fail. The right fix for this is just use cp -dpR (actually it's -dpPR, but -P is not POSIX, so..). I haven't listed them for now, but it's still a problem.
Well.. time to see how many other packages I can fix before going to my LUG for the meeting.
16 May , 2005
The most common errors during gcc upgrade
Ok second blog entry of the day, mainly because I don't want to go sleep and I have nothing to do which doesn't involve be completely awake.
News for who care: I joined pam herd to help az with pam modules and configuration files handling. That's related to the work I'm doing to have pam working out of the box on Gentoo/FreeBSD, which requires to change a few things here and there.
Also, if you haven't read the comment to my previous blog entry, you can find on Planet KDE, a rule to filter out the german spam in Dirk Mueller's entry. Thanks Dirk for that :)
Now the topic of the entry... after release of gcc4, I started caring of portability fixes to have GCC4 working for some video stuff which broke. That's common in video and sound software because to achieve performance, it's not uncommon for programmers to use dirt hacks and trickies to carry out some processing. That, unfortunately, often makes programmers "assume" that their system is the same of every other. They can't be more wrong.
Let's see the most common errors which throws up when a new gcc is released and it gets more strict.
The first one is the declaration of static vars. Many developers just defines variables in .h and use them in just a .c source file as statics.. that's not good because the .h defaults to extern. Another problem is when a variable is declared static and, with the same name, is declared extern elsewhere (see transcode problems)... that's bad.
Other problems can be with some strange pointer arithmetics, which gets fooled up (xine-lib had this problem and required a couple of patches to get rid of it in the right way, as the more logical way from a non-expert pov is wrong by practical check). You should nevel try to write something in a way another developer can't read, because probably neither newer compilers can read it in future. An assignment more shouldn't be so much delay in the code.
Finally, there are my preferred errors: pointer foolups. GCC is getting more and more strict in latest version about it, and that's good for everyone in a 64-bit architecture. Trying to summarize, many people just think that pointers are always 32-bit and they are always of the same size of int integers. Nothing can be more different.
On 64-bit systems, the pointers are sized... 64-bit, but int integers are just sized 32-bit as usual. Instead long integers are sized 64-bit, so long is the right type which is of the same size of pointers in current systems and compilers.
There are many ways to be sure that you don't use the wrong type for those things.. for example you should never ever try to shade an int into a void* pointer... you should neither try to store a pointer into an int, if you really need to store a pointer use ptrdiff_t which is guarranteed to be of the same size of pointers (if using gcc or gcc-compatible compiler) or at least a long integer.
So in general, the new warnings in gcc 3.4, which turned into errors in 4, are good things (tm) because they allows 64-bit systems' users to be sure that the program they are running doesn't try to do foolish things which causes segfaults and so on...
Now... you should think that something like that came up only on some strange multimedia stuff or in software which is poorly developed or not widely used.. that's also wrong.
Following Mark's entry I tried to build mozilla-firefox with gcc4, because it's way and way faster.. unfortunately it fails also if gcc4 patch is applied, because firefox's code suffer from many of the coding error I've just explained: integers shaded into void* (which breaks when you de-shade them), pointers stored as integers, and so on.
I don't know how could Mozilla work fine on 64-bit systems with such problems, as they tend to corrupt the memory and makes the program segfault. But it's true I usually use Konqueror and not Firefox so it can probably segfault often. Also many users just use 32-bit firefox to have Macromedia Flash support so they can't see the problems.
Well... I think I'll try to prepare some experimental ebuilds which uses kde split ebuilds and svn modules to try to build them with gcc4 and fix the problems where they aren't fixed in svn (sheees.. I was writing cvs... I still haven't made up my mind about kde's svn move!).
15 May , 2005
Fewer patches, fewer combos
After the new warning of the last days about toolchain-funcs eclass (well actually about gcc eclass which is getting deprecated), I started fixing all the video and sound ebuilds to use the new toolchain-funcs eclass, where needed.
Yes because many ebuilds stopped using gcc functions but still inherited gcc, or where just using it to find out if you are using gcc 3.4 to apply patches which are valid for all gcc version. In these cases, I just removed the inheritance and made the patch apply unconditionally.
This is more important as it seems, because sometimes the new errors which came up with newer gcc was just warnings before, but the code was equally wrong. This is true also for new gcc4 patches. Having the patches applied unconditionally, also, makes less combination of applied patches possible. This is useful because if the code is different between system cause of the gcc version, it's more difficult find out problems, also because emerge info just shows the current compiler, not the compiler used for the package (this is true mainly for runtime problems, not compile time as there the gcc is just the same).
Cleaning up many ebuilds also lead me to check some quite-unused quite-unmaintained packages, which shown 34K patches and 22K makefiles. It's not so much, but 120K less to sync is someway interesting for me. I've already converted a few packages to patchset tarballs (that I maintain locally using a subversion repository, so to have them always updated), but those was less band-consuming that the ones I found last night.
With that cleanup my commit statistics are going to have a strange curve :)
Now I'm compiling gcc 3.3, to test a couple of packages so to apply a few more patches unconditionally, so to have less checks here and there.
I've also changed the checks for gcc 3.4 or better, when legit (for example for cflags), to consider the existance of gcc4. Most of them was just bound to gcc 3.x series.
By the way.. have you seen the german spam attack? It's driving me crazy, also because usually most of the spam is just filtered by GMail, and a few mails are filtered locally by spamassassin.. today I received more than 60 spam mails which was unfiltered, and just now spamassassin starts to recognize them as true spam. As suggested by slarti I've increased the scoring of razor2 and bayesian rules so that they can junk a single mail by themselves, and now I hope I can get rid of them.
I'm lucky I'm using a good-trained bayesian filter, but that's not a solution.
13 May , 2005
Ready Kernel Go!
Ok a bit of a citation of one of my most loved songs (L'Arc~en~Ciel - READY STEADY GO!, second opening for Full Metal Alchemist anime) for a good, or better than good, news.
It's been a week since my last blog entry, mainly because I was a bit occupied with work, real life and Gentoo itself, but also because I wasn't having news about G/FBSD.
I'm waiting a patch for binutils by SpanKY/Vapier this weekend, so to start messing around and having xorg working on G/FBSD.
In the mean time, FreeBSD project released version 5.4 final, or 5.4-RELEASE to use their own name. I have updated all the ebuilds from 5.4_rc4 to that version, as they needed just a rename. But I haven't tried upgrading the kernel until today.
I wanted not to try compiling the kernel with release candidates because that could have been unstable, and I risked to have a non-working system because it has no monitor attached to it and if the kernel failed I had to take the old one which is currently in another room.
Today, I wanted to try it out, and I took FreeBSD handbook to look how to compile the kernel and then compiled it with a custom configuration file. It worked at the first try, after fixing a couple of problems with awk/gawk (still incomplete the fix, I know), with libl/libfl mismatch and with -Werror (gentoo's GCC is a newer version with backports, so it has more warnings). Lucky me:
flame@defiant ~ $ uname -a
FreeBSD defiant 5.4-RELEASE FreeBSD 5.4-RELEASE #0: Fri May 13 12:13:33 UTC 2005 flame@defiant:/usr/src/sys/i386/compile/DEFIANT i386
What does this mean? This mean that we can install a full system using portage and also compile the kernel after having it installed from the ebuild.
Also, we can apply the security patches from FreeBSD project just bumping the revision and install a new copy of the sources. I still need to find a way to add the symlink useflag support to freebsd-sources, but I'll find a way sooner or later. Another solution could be to prepare n eclectic module to handle it :)
Well now what's the main problem? I haven't found how to install base files for boot0 (freebsd's bootmanager), vapier needs to send me the patch for binutils, and we need to find how and whether to install the documentation which is not handled by portage on my system.
My hopes now are to provide an experimental stage for installing next week. That will require use of a FreeSBIE livecd because we haven't started on working on a Gentoo/FreeBSD livecd for now.
In the mean time.. we are plenty of ebuilds in overlay to merge in main portage and a lot of ebuilds needs to be fixed for pam support (I'm doing that right now).
7 May , 2005
Another day with less bugs
Okay another day as gentoo dev which involves fixing bugs and enhancing Gentoo/FreeBSD.
Last night I started looking at video bugs one by one to close all the ones which were fixed by other bugs, the ones which are probably fixed in newer versions, the ones which wasn't active in the last months. Now we are at a human number under the 90.
Now I'm in my garden, under a few trees that I complete an upgrade to enterprise (the linux box) and updating the cvs in defiant (the fbsd box) to see if I'm able to merge more patches inside main tree.
Just to update the situation with G/FBSD and Xorg, I'm still getting problems with opengl linking. Pam stuff should be fixed also if I can't test and I'm not receiving news from fd.o bugzilla about it (well I've reported it just a few days ago, it will take a bit before someone can look at it I think). OpenGL problem seems to be related to binutils which ignores ld.so.conf file and fails to find the libstdc++ file where it should. I've found a patch to binutils on Debian GNU/kFreeBSD Project after spyderous pointed me that they had the same problem. Now I only need to find the patches to apply to binutils to support FreeBSD systems as well as linux systems in gentoo.
6 May , 2005
G/Fbsd status update
The last two days were so intense that I couldn't manage to post something here. Yesterday I had just a few hours of time to spend on gentoo actually, mainly because I had a friend of mine who came visiting me after something like six months and then there was the meeting of my lug which took me quite all the night.
This doesn't mean that G/Fbsd is frozen or something like that. Actually, yesterday was also released 5.4 release candidate 4 of FreeBSD which I'm trying to prepare for G/Fbsd project ASAP. Actually, I hadn't updated everything to 5.4 rc3 and moving to rc4 needed to redo some changes I was doing to rc3. And, also if I was able to build for the first time Xorg, the second compilation didn't go so well: there are features which are activable just on the second compilation, and some of them are broken, for example pam support doesn't work with OpenPam, the PAM implementation used by FreeBSD, and Xv/OpenGL support doesn't seems to be able to be removed for now. I've managed to fix the PAM problem (I think) but I'm getting a couple of problems during glxinfo linking with opengl flag enabled. I spent two days to go to that point and now I need to find the solution in the next days at every needed cost.
Well anyway today I started committing rc4 ebuilds, after having prepared a script which I used to download the source tarball fragments from FreeBSD's ftpsite and repackage them as tar.bz2. I've also uploaded to mirrors all the rc4 files so that they are available for testing, but remember that 5.4 profile is still *a lot* incomplete.
During the merge of newly system's ebuild I found out that using portage's flex breaks compilation of quite every fbsd's package, that's because it tries to link to libl but in flex is called libfl. A few patches fixed the problem.
Now I'm having a bit of a problem because I removed a library which wasn't mean to be removed and I need to find a way to have it working again, I hope to fix it tomorrow and then commit also usbin ebuild.
News from the 5.4 profile includes for example the complete drop of vixie cron from within freebsd-usbin: now if you want it you just need to emerge vixie-cron as in Linux. That makes the packages lighter and also saves us from supporting all the init.d files for software which is already in portage of its own. This is true not only for cron (which wasn't being disabled by a useflag at all), but also for sendmail and bind (which had a useflag before to enable them): now they are just not built at all and the useflags are gone: if you want them, just emerge them separately :)
Please note that making software not installed by freebsd-* packages doesn't mean that we are going to replace it in the base system, as we aren't replacing them with other default software (also if, in case of patch, this could have been done), just we replace them with different versions of the same softwares. This makes 5.4 profile less adherent to freebsd layout but more adherent to gentoo's one, so installing it over a minimal installation won't remove the outdated libraries as 5.3 used to. To fix this we need to find a good script which removes orphan files, I think findcruft could do the job, but we need gfind to have it working. I'll work on this tomorrow.
Also, as 5.4 goes final and the profile is safe, we're going to move all the init scripts for software provided by freebsd-* packages directly into their ebuild, moving them out of baselayout, this should simplify the problems to Gentoo/kFreeBSD project, I think.
After completing 5.4 passage, I hope we can be able to prepare a stage3 to install g/fbsd without need to install before the clean fbsd. But this is not something to plan for now.
Tomorrow I have a lot of things to do, I hope to find the time for all of them.
3 May , 2005
And finally, the graphiX
Another status update for the Gentoo/FreeBSD project. It's getting a daily job do that, but this time I have great news and a screenshot!
Last night I was finally able to build xorg-x11-6.8.99.3 on the g/fbsd system! It has no opengl or xv support for now as it fails to link to libGLU and I don't know why for now, seems like it misses to tell the linker where to find libstdc++ library. But a part from that is a completely usable X environment.
Actually you can't build it right now completely as it tries to install xterm which depends on utempter which doesn't work on non-linux systems (praise RedHat for that as utempter is their stuff); instead there's libutempter, a drop-in replacement developed by altlinux which works fine both on linux and fbsd; seemant will take care of libutempter and xterm this week so soon xorg will build out of the box.
My project was to build fluxbox and use it to try some xish stuff until gcc problems faded out and we can move to complete the 5.4 profile for the first stage release. Unfortunately, fluxbox doesn't realize correctly when iconv is installed or not, so I think I'll need to prepare an m4 library file which checks for iconv support to propone to fluxbox (upstream and gentoo's) so that libiconv is corretly found where it should.
Until that, I'm going to work on ghostscript so to have a working version for g/fbsb which is needed by kde to build. Really if we are going to have a working kde environment in g/fbsd soon, we can compete with that project which is going to prepare a kde-based bsd distribution.
Really would you imagine the power of FreeBSD combined with Gentoo and UIs like Gnome and KDE like they are on G/Linux? :)
2 May , 2005
G/FBSD and X.org
Ok waiting for problems with $LIBC variable in portage get fixed (for anyone hasn't synced and tried to compile python yesterday, the compilation was broken by an "unknown file" glibc which was caused by the new global $LIBC variable which was being used by python whithout sanity checks to see if that's really the var it was searching) I'm still working on G/FBSD trying to get latest xorg (6.8.99.3) working fine on that system.
Actually, my g/fbsd box (defiant from now on) is an headless system, I have no monitor and no keyboard attached to it. What I want to be able to do is just having a vnc session opened to it so that I can test X-based things. My dream is getting KDE running on G/FBSD with the less modifications needed.
Well now I'm waiting for compilation results from xorg-x11 emerge process, still out in the garden with my cat laid at my side. The cause of the recent failures in compiling xorg was found in a single patch which was breaking compatibility with non-glibc systems and for that wasn't accepted in main xorg tree. I hope that excludign it when libc is not glibc is enough to get this working.
So do we have news on G/FBSD project? spb is taking care right now of updating project page, and today's portage was broke. But that's going to be fixed soon, I hope. Now we are mainly waiting for LIBC, KERNEL and USERLAND expands to fix dependencies over libiconv and gettext, and get finally a working out-of-the-box irssi, without overlaid ebuilds.
After that there are a few points which needs to be fixed before getting into trying the first release, for example kernel installation, boot loader setup (grub?) and obviously documentation, althoug the combined gentoo+freebsd handbooks should be enough for start with it :)
Time to get into house, here the bugs are ating me!
1 May , 2005
When your upstream need patching
Ok today here in Italy is holiday and I stopped working on the translation I'm paid for (it's also sunday so it's "double holiday" :) ).
The day is good and the sun is shining so I'm out on my garden with my iBook working on gentoo via ssh to the main box and screen. X-Chat Aqua let me maintain the link with #gentoo-dev and other chans.
The cell phone is off an I feel more free than usual. So I'm in a good mood for a blog entry :)
One of the most annoying task when maintaining packages on gentoo is maintain clean the patches which, by version to version, doesn't apply and needs to be re-created mainly manually. This task is needed because sometime we have to change the way packages works to make them follow gentoo's structure in installing files or other things.
Other usual patches you need to prepare are when upstream didn't cared about 64-bit architectures (which were pretty rare until a few monhts ago we can say, as amd64 (Athlon64 and Opterons), ppc64 (G5) and ia64 (Itanium) are recent processors. In those case you can have a lot of warnings about int casted to pointers and the other way 'round. The problem here is that the size of a pointer in a 64-bit arch is 64-bit instead the size of int is just 32-bit. long type is usually the same size of the pointers, and ptrdiff_t should be used to do arithmetic computation with pointers.
Also with the newest gcc (4.0.0) most of those warnings are became erros and need a complete fix. This is good as code ingoring those warnings usually produce wrong binaries.
Another kind of problem is when borked configures doesn't like passing --enable to commandline (for example when they consider every --enable|disable-something as a --disable) or when they just use automagical test for libraries presence, which makes useflag not honourable.
In video category there are many apps which has a lot of dependencies in external libraries to mangle file formats, and many of them doesn't provide enough configure params to satisfy our useflags needs, so I have hacked most of those configure scripts to add params and sent them upstream. Most of them was applied upstream, a few weren't and I'm still hoping for them being applied as they are a big pain.
One of the package for which I'm having most patches is xine-lib. Mainly because it's a popular package which is installed in a lot of different machines with different setups, but also because its configure.ac was really messy, and for a long time the package wasn't able to exclude some plugins and the ebuild used to depend on something line aalib which was useless for most users. Then there was my rewrite which used autoconf cache tricks, but it was probably not working with confcache so I preferred to patch xine-lib to add the needed params. Still the patch is not applied and it needs a lot more work to have it clean anyway.
But it's onot only the configures which are a pain to fix because they aren't complete or just suits the setups of the developers, there is also the PIC __PIC__ stuff: libtool used to define -DPIC to select between PIC and non-PIC stuff, but gcc already defines __PIC__ if -fPIC is passed, and sometimes the -DPIC is simply missing for some reasons and the things need to be changed accordlinh.
Most of those errors are things which gots warning or which needs to pay attention to to be sure that they works, but they can be real pain when the complier start to be stricter (which occours every gcc release).
Now, I hope the new patches I'm working on for xine-lib can be applied upstream so next release would simply drop the patches.. it's being a few versions in which the patches aren't being dropped but added (well, a part from the security fixes and the wma stuff added in -r2, -r4 and removed in 1.0.1.
Oh I forgot to say what I'm working on at the moment... it's just a little little patch to enable vidix support on amd64 (I had a report for a gcc4 problem in vidix driver, I fixed it and tried to compile it but vidix support wasn't being built at all, then I started looking at it and I found out that it wasn't supported on amd64, and I've make it be accepted and compile. Now the vidix output plugin is on my system, but it seems not to work at runtime. I'm investigating this.
Oh you'll find that post in an hour not suited to be outside in my timezone (20:30), but that's because I had a friend here and stopped blogging.
Diego Petten
| Mon | Tue | Wed | Thu | Fri | Sat | Sun |
|---|---|---|---|---|---|---|
| < | Current | > >> | ||||
| 1 | ||||||
| 2 | 3 | 4 | 5 | 6 | 7 | 8 |
| 9 | 10 | 11 | 12 | 13 | 14 | 15 |
| 16 | 17 | 18 | 19 | 20 | 21 | 22 |
| 23 | 24 | 25 | 26 | 27 | 28 | 29 |
| 30 | 31 | |||||
Search
Categories
- All
- Gentoo (161)
- Gentoo/*BSD (91)
- KDE (45)
- Personal (88)
Archives
- April 2006 (17)
- March 2006 (19)
- February 2006 (38)
- January 2006 (35)
- December 2005 (36)
- November 2005 (32)
- October 2005 (44)
- September 2005 (26)
- August 2005 (20)
- July 2005 (24)
- June 2005 (18)
- May 2005 (11)
- More...
Linkblog
Gentoo
- More sites
Its been crazy the last few days with Gentoo's infra. I helped setup this planet site for dsd over the weekend and will be released in a few days. So far it seems to be working great! The next site I've been helping bring to life is the scripts repository site. This site will help bring together any scripts that people have created for Gentoo. Ian Leitch has been great to work with to get this site up and running. Now he'll finally be able to test it in a better format :)
Another project I worked on lately was helping setup a box for Brian Harring for the xdelta project he's working on. He'll have a server all to himself to torture and see how things go.
- Recent migrations galore
Its been crazy lately with all the service migrations for Gentoo infrastructure. I'm just glad that most of gone smoothly! I'll be glad when we get all the services off of eagle so we can finally move that server to its new rack. Finally got around to getting Planet Gentoo setup for dsd and it looks sweet! I can't wait for us to nail any issues with that and and have our users start using it. It'll be a great addition to Gentoo for sure.
Gentoo
- /dev/urandom thoughts.
On Saturday I visited the folks at Salford uni to attend the Gentoo UK 2005 Conference. There is a fine write-up on it in this weeks GWN so I won't elaborate on this too much, but I would like to extend my thanks to all of those participating in the event this year. It was a pleasure to meet those dev's I've never met before in person. Shouts out to Tim, Tom, Dan, Stuart, Rob, Stephen, and although I never recognized you on the day Marcus! If there is anyone I have forgotten, my apologies and shouts to you too!
Gareth Bult of Flash Linux fame spoke about the technical limitations of USB keys, which I found most interesting, and also (indirectly) raised a few points which I would like to rant about. Documentation! Everyone knows our documentation team do a great job and our handbooks are nothing short of superb, however there are so many other documents which we look after which are terribly outdated or have not been made aware of. Hopefully the planet is a good push towards the aggregation of information, although I for one will be making more of an effort to keep documentation well organized and up to date. Daniel Drake (dsd) spoke about his views of the kernel, mostly the 2.6 branch and its organization and touched on a few nice subjects. Monolithic vs. Modular for example. I felt a little embarrassed that I attended and didn't put in any talks of my own so I must apologize for that, however I thoroughly enjoyed Dan's talk and he would have shown me up anyway ;) Something I would like to add however is that in the coming few months I am going to make a more conscious effort to keep the project page updated and our TLP roadmap accessible. With 2005.0 still being up-in-the-air I am going to hold off however. Unfortunately I missed most other peoples talks in full as Stuart and I ran off to the side-room together! But from what I hear Rob only swore once, so way to go! All in all, thoroughly enjoyable.
On a different note I went to Alton Towers on Friday and even the weather held out! It was a lovely day, and it was an awesome amount of fun. Anyone who's going, I recommend staying the night in "The Bulls Head Inn" its just down the road, and the breakfast is fantastic. I think I went on every ride coming close to 4 times or so. Hex was the biggest dissapointment but numerous goes on Oblivion and Nemesis made up for it :)
Gentoo wise, there are several things coming up in the next few weeks with Kernel. There is of course the 2005.0 release which has been prepped for and requires further work once released to clean up old packages in the tree and so on. There has been some excellent progress made in migrating all the older sources to kernel-2 and older kernel module ebuilds to linux-info/mod eclasses. I will also be auditing our version detection mechanisms in the eclasses to ensure the recent move to a more refined upstream release scheme will be sanely catered for, and also addressing any issues which may have popped up from my recent unipatch change. Which reminds me, I am actually going to finish that re-write soon so devs can expect a much more powerful unipatch syntax and speed-ups. I would also like to welcome Carlos Silva (r3pek) on board! It's going to be a pleasure working with you.
So there is my first ever blog post! And I would just like to take this opportunity to thank Dan and all else involved for their dedication and initiative which made Planet Gentoo. It truly is an excellent tool!
- Asus Pundit-R's, Asterisk and Kids on Bikes
So its been a little while since I last posted so let me update you all.
My Girlfriend (Claire) and I are looking around for a house, making the big move in together. I never realised how stressful just looking is! We have seen a fair few that we like, and have arranged several viewings but time will tell. I've also got quite addicted to "Ladette to Lady" on TV. I didnt realise watching stupid pompous old grannies and crazy young girls would be so entertaining.
Oh, and then there is my car. The accident magnet. As some of you probably know some stupid woman crashed into it, which I had to claim for an so on, and I have just now (after months of waiting) recieved the estimates. Well, I sat down for my dinner the other day and the door-bell rang so I went to see who it was. Some kid (good on him for not running off mind) appologised for riding down the road, losing control and crashing into the side of my car. It left a rather tidy scratch all down the rear passenger-side panel, and also a nice dint. Less than impressed :(
Also, no idea how many people have seen this but its pretty awesome. Basically, 18 real life taxi cabs fitted with GPS and split into teams of three. You pick a "team" as your online monopoly piece and when a cab is near/on your property after the round is up, you get paid rent. equally you pay rent in the same way. Very cool!
Anyways, on a more technical note I've been playing with the Asus PUNDIT-R's as a solution to running Asterisk with some difficulty. The digium card (TE110P) is based on a well documented, open card with open specs. Problem being there is just enough variation in it to make it a pig. Once you enable the spans on the card, the card will begin to send interrupts (in a frequency similar to the timer) and also enables DMA access. now, the IDE bus on this machine has a faulty DMA as it is, and also it appears a faulty (IO/L)APIC implementation.
Im still in the process of trying to diagnose as to why the box will hardlock under minimal load exactly, but it is almost certainly to do with the way it handles DMA, and more than likely it just clobbers userspace memory regions which will then be over-written by userspace, which then currupts kernel-space and hangs.
However, if anyone has any experience with these boxes, this hardware, and asterisk please give me a shout and let me know how you got on. I have even tried forcing interrupt allocation to the BIOS in a check to ensure sensible sharing. - death to modconf
for those faithful following my heartbreaking drama story of a car and its owner, there is still no progress been made. The weather is getting wetter, and my poor baby is trying to hold the fort against the elements to prevent itself from rusting, and although I fret I have began to come to terms. Still no news about claiming for its repair yet, and still no news about making a statement but I suppose thats just slack police :)
A few things happening in gentoo land.
modconf has been removed, excellent. Its been in the tree (same ebuild, only trivial changes) for 2 years. It had come to the decision of keeping it, and bumping it to working or dropping it. After brief discussion, the latter prevailed.bugs #85410 and #84856 are closed. Anyone having problems with unipatch working on something other than base10, and madwifi not building if you use KBUILD_OUTPUT things are looking up! :)
bug #77190 has been closed. Anyone who was setting a LANG/LC_ALL variable which screwed up unipatch should now be working fine without needed to mess with anything.
And, plenty more to come. All in all, I don't have a great deal to add really. Only thing worth noting is I'm not feeling well and if things get much worse my availablity might become a little awkward.
- Ex-employers can really suck!
So, all in all this has been a fun weekend. The weather has held out which is good, I have a new car (new Hyundai coupe UK US: works under epiphany!) which I've been driving around a lot all week.
I've been on the phone every day to Manx Telecom (my ex-employers) recently trying to arrange for my internet access to be reconnected. One of the perks of working there was free ADSL, however for some anomaly it was never added to my line. Therefore, it was ceased and I have had no internet access for almost a week. Apologies to those waiting on me for stuff with Gentoo, but the above explains my lack of activity this past week :)
I've also been dabbling a lot recently in the new multisync cvs builds, uclinux updates and a couple of other goodies. Hope to push some of it to the blog/tree soon. On top of this I'm going to commit nicer support within detect_version for the newer kernel scheme, something I've wanted to do but with 2005.0 and my lack of net access its had to wait.
- Kernel Sources
For all of those awaiting a more permenant fix to bug #85559, this has now been done. Hopefully you vanilla-sources users (specifically) will benefit from a big bandwidth saving.
Also on a similar note, there has been a lot of confusion recently about 2.4/2.6 kernel versions and headers. Let me clear this up.
Many moons ago portage didnt have support for cascading profiles, although the 2.5 kernel had just been made 2.6 and progress was being made on stabalising support for it in Gentoo. The issues we had meant that we had to rename the 2.6 versions into a new package. For example: linux-headers contained 2.4, and linux26-headers contained 2.6.
This meant that managing the dependancies within ebuilds was awkward and amongst other things, far from ideal.
It was also an illogical seperation of what is fundementally the same thing. You dont for example see vim5 vim6 etc, you just have vim.Now then, what we did recently, with the help of cascading profiles was amalgamate these packages into their relevant counter-parts. Therefore, we now have vanilla-sources-2.{0,2,4,6}* and linux-headers-2.{4,6}* and it is up to the profiles you run to manage which versions should be unmasked for you.
As part of this move we also moved to 2.6 by default for many architectures. As a result, and in true gentoo philosophy, you will find underneath your profile either a 2.6 or most likely a 2.4 subdirectory. If you link your profile to that directory instead then you will no longer be forced to update to 2.6, however I do encourage you to upgrade if you have no valid technical reason to stay.So with this concludes:
emerge yourfavourite-sourceswill emerge 2.4, OR 2.6 depending on your profile. Most likely 2.6
emerge linux-headerswill merge the appropriate headers.IF you are upgrading from 2.4 to the newer 2.6 as part of this move, PLEASE PLEASE ensure your new kernel is installed and running along side your new 2.6 headers, since there are several reports of random segfaults occuring with 2.6 headers on a 2.4 kernel.
If you find that its installing a version you dont want, then just relink your /etc/make.profile to ${PORTDIR}/profiles/default-linux/x86/2005.0/XX where XX is 2.4 (or 2.6 on different archs in some cases).
Hopefully this has now brought some clarity to the situation :)
- Stupid French Cars!!
So shortly following the purchase of my new car, I was driving home at a very reasonable speed, when all of a sudden a newly passed driver in a citroen ax came around the blind corner too fast hitting the car in front of me. So, I swerved to not get hit by the spinning AX, and bits of the cars were flying all over my bonet.
I rang 999, done the normal stuff - luckily everyone was completely fine. Anyways, checking the damage to my car and it was nothing worth crying over I left and went home. While at home I saw that it had ripped big chunks out of my paintwork all over my bonet, door panels and bumpers.
After spending a good half an hour on the phone to a police officer dealing with the accident, I think he finally believed me and so I took it to the local station so that they could check it. Now all I need to wait for is something to happen to pay for the damage to be repaired before it starts to rust!And to add to the annoyance, the only reason I drove away from home in the first place was to pick something up from a shop which rang me to say something I wanted was in, only to find by the time I got there they were mistaken!
So, anyways, Gentoo stuffs.
kernel-2 changes have gone in to better accomodate KV_EXTRA and family.
linux-mod changes have gone into the tree to take over the pcmcia work from pcmcia.eclass, and pcmcia-cs changes will be made soon.instead of it now working out and patching a load of odd pcmcia sources, it just tarballs up the pcmcia-cs sources at build time, and uses that for the future. Please please please dont delete /usr/src/pcmcia/pcmcia-cs-build-env.tbz2 once these changes go in or you might experience problems :)
Aside from that, nothing new to report.
- tasting that fresh mono goodness.
So its been a while since I last blogged, and I've decided to give in on that whole "I promise to blog more often" routine which just doesn't work, but after having a few things happen recently which someone might actually like to read about, I decided to write a new installment of my crazed thoughts to entertain those religeous few :)
I've been looking for a simplistic, yet powerful Podcast client for quite some time now, without any of the ones i've found (iPodder/Juice, Rhythmbox etc) being simple and specific enough. I fairly recently came across monopod which I wrote an ebuild for (0.3) and after finding a bug open for it on bugzilla, submitted it to portage.
At the same time, I decided to clean up v0.4 and got right into mono development. So far I've fixed up the deprecated code, fixed and partially re-worked the iPod support, cleaned up a lot of smaller UI niggles and started writing a plugin system fairly similar to Banshee's to support automatic sync to iPod, daap, etc etc.
I've been in touch as well with Edd Dumbill and hope to start putting more time into turning monopod into a very convenient lightweight, but extensible podcast client. Of course, the fact that Banshee (which is awesome by the way, thanks Aaron) is actually getting a lot of attention from people writing podcast plugins for it means that monopod might end up being fairly short-lived. But obviously it has its purpose and I would never encourage playback support in it by standard anyways.
Anyways, on a totally different note Tim (Plasmaroo) lisa (lisa - funnily enough) and I met up in York for a bit of a gentoo get-together with a few other people on Saturday. It's nice to catch up with people face to face, and Tim's ability to shout russian in Pizza Hut impressed me! We met a rather interesting poet in the bookstore and ended up chatting about the ups and downs of (iirc) Jasper, XML, XSLT, Why not to use JavaScript, and then participating in some amateur filmography at the top of the stairs! :)
It was fun, hope to do it again sometime. The opportunity will come sooner than expected too with an unofficial meet in manchester shortly and a Gentoo UK gathering planned sometime near late May/June in London. Of course, everyone will be welcome and all interested parties should express their interest by badgering George (cokehabit) on #gentoo-uk ;) - I'm curious about rough numbers as I'm sure George is as well.
So, I could go on for a while with all the things I've been working on recently, but instead I'll give it a break and leave some beef for the next few days :)
Also to note, David Nielsen (Lovechild, some of you may remember him from his gentoo days) has been sexually abusing a lot of the UK developers recently. Word of warning for those tempted to visit us in London ;)
Gentoo
- Dual Core G5s (970MP)
Looks like dual core G5s aren't that far off, if you take the update to MONster to be any indication. If you all remember last year the 970FX definition showed up all of 3 months before the machines hit the shelves. Apple has a tendency to only do major product releases three times a year, Mac World Expo in San Francisco, WWDC and Mac World Expo Paris. If the past is any indication of future results it looks like they are trying to push for production machines by WWDC in June. With the recent updates to the ppc64 kernel, and new fun stuff like AGP and iMac-G5 patches coming down the pike it looks like ppc64 is going to grow fast from here on out. Now if I could only get multilib working...
- Hardened coming to ppc64
Just a heads up, I'm working to bring the Gentoo hardened profile to a ppc64 near you. A big thanks to solar for putting in the time to help me with this. I now return you to your regularly scheduled programing.
Some preliminary PaXtest data (no toolchain or noexec/pageexec yet):Mode: blackhat
Linux Strife64 2.6.11-hardened-r1 #4 SMP Wed Mar 16 21:08:23 EST 2005 ppc64 PPC970, altivec supported PowerMac7,2 GNU/LinuxExecutable anonymous mapping : Killed
Executable bss : Killed
Executable data : Killed
Executable heap : Killed
Executable stack : Killed
Executable anonymous mapping (mprotect) : Killed
Executable bss (mprotect) : Killed
Executable data (mprotect) : Killed
Executable heap (mprotect) : Killed
Executable stack (mprotect) : Killed
Executable shared library bss (mprotect) : Killed
Executable shared library data (mprotect): Killed
Writable text segments : Vulnerable
Anonymous mapping randomisation test : 24 bits (guessed)
Heap randomisation test (ET_EXEC) : 14 bits (guessed)
Heap randomisation test (ET_DYN) : 32 bits (guessed)
Main executable randomisation (ET_EXEC) : 20 bits (guessed)
Main executable randomisation (ET_DYN) : No randomisation
Shared library randomisation test : 24 bits (guessed)
Stack randomisation test (SEGMEXEC) : 32 bits (guessed)
Stack randomisation test (PAGEEXEC) : 32 bits (guessed)
Return to function (strcpy) : paxtest: bad luck, try different compiler options.
Return to function (memcpy) : Killed
Return to function (strcpy, RANDEXEC) : paxtest: bad luck, try different compiler options.
Return to function (memcpy, RANDEXEC) : Killed
Executable shared library bss : Killed
Executable shared library data : Killed - Nerd Score
Yeah, even though I'm on vacation I just had to jump on the band wagon. Damn peer preasure........
- PowerPC to the People
10 PRINT Hello_World
20 BEEP
30 GOTO 10
Ah gotta love Apple Basic.A little story for introduction:
At the edge of the Architecture map the intrepid programmer found the words "Here there be PowerPCs". Having no fear of these mysterious processors he set his sails to catch the wind and found that indeed the world was not flat. What he found over the horizon was a land where code was no longer bound by the tyranny of x86, a veritable paradise. The programmer set up shop and hung a sign outside his door; "PowerPC to the People" it read. As people slowly realized there was another way they broke free from their shackles and came to the new land. Welcome the programmer said, stay a while.
- The 'What we did in 2005' bandwagon
Ok, so jumping on the trend started by Simon and Diego here is the 'What did ppc and ppc64 do in 2005?' status update.
- The first thing to mention is 2 very successful releases each adding futher support for the machines using the powerpc processor. 2005.0 and 2005.1 were both successful. Additionally 2005.1-r1 fixed some minor issues on PPC64. Thanks got to Pylon, wolf31o2, jforman and the entire PPC, PPC64, RelEng and Infra teams.
- Along with 2005.1 we merged the ppc and ppc64 profiles into one common parent to better match the efforts of sparc and mips which both support similar structures. The merge was mostly stylistic but a further blending will be coming with 2006.0
- Support for the PPC970 processor found in the G5 transitioned over to the ppc64 team. Needing an easy way to transition users we created pure 32-bit, pure 64-bit and multilib userland profiles (the latter thanks in large part to the AMD64 team whos work made it possible). Gentoo is the only distro out there that fully supports all three types of installs. That is huge.
- 2005 saw the first support for Gnome in a pure 64-bit environment on PPC64 as we finally got mozilla to compile. Mozilla and Firefox still don't work but efforts are continuing to make these browsers ppc64 64-bit friendly.
- Hardened support has improved on ppc and ppc64 saw the first hardened profile. Neither one is really ready for prime time, but thanks to the great work of our Hardened team things are getting there.
- We started using ATs on the ppc team, they have been a great help, thanks all of you and thanks to the AMD64 team for coming up with the idea.
- Due to all these improvements, and the continued improvement of the PPC Faq and the Handbook we were able to close just shy of 950 bugs between the two groups.
- Because we have made such a name for ourselves as a strong reliable distro for both ppc and ppc64 we saw huge hardware sponsorship from Genesi and IBM. This relationship I'm sure will continue to grow. Thanks go out to both companies for their continued support.
- While not technically a Gentoo accomplishment work continues on the bcm43xx driver for the AirportExtreme (among others). I would personally like to thank JoseJX and Kugelfang for their contributions in bringing this project to where it is today. There is still work to be done for sure, but hell, I have wireless on my iBook now so I can't complain. I'd also like to thank all those who work on the driver that are not directly part of the Gentoo community, good job guys!
- I'd also like to thank all those that work on the ppc32 and ppc64 kernels as 2005 saw support for quite a bit of new hardware and without them it would not be possible.
- Finally I'll leave off quoting Simon, as it truely is the most important aspect of all: "We had lots of fun".
All told I'd say that's one hell of a year, here is to another great year for Gentoo, the PPC architecture and OpenSource as a whole.
Gentoo
- Filtering TOFU
This morning I discovered
net-mail/t-prot. It's specifically designed for mutt users, but it should work with other MUAs, providing they're not one of these new fangled bloated graphical things.Anyway, here's a URL: URL
The idea behind it was originally just to filter out classic TOFU, that is, "text oben, full-quote unten". This is a mish-mash of German and English meaning "text above -- full quote below", or just top posting to the rest of us.
However, t-prot filters out more than just TOFU. It gets rid of Outlook garbage and it can trim commercial and mailing list footers (or whatever footer you like). It can truncate RFC uncomformant signatures that are over four lines long. It does a bunch of other things too: trimming whitespace, repeated punctuation, blank lines, etc.
The best bit is that because it's just used as a
display_filterin mutt, the original mail is unchanged. This means there're no strings attached, so try it out.Just for the hell of it, here's a screenshot before (left) and after (right). Click on the images for full-size, if you're bored.
Interestingly, the person having their mail snipped by t-prot for having a huge RFC unconformant signature is also part of the ASCII ribbon campaign. It takes all sorts, I suppose.
- Gentoo UK Conference 2005
Just got back home after my flight back from Manchester. I'm very tired, but I'll do my best to scribble down a few things. I apologise for not having any photographs, but there is a video/DVD in the pipeline.
Rob Holland (tigger^) gave a great talk on code auditing, in particular with doxygen and his work with that. The slides were a bit rough and ready (hehe), but it was excellently presented nonetheless. He didn't even swear once.
Stephen Bennett (spb) showed me and a few other people Gentoo/FreeBSD with the Gentoo init script system. Really quite impressive.
Daniel Drake (dsd) presented the kernel and user-relations projects. I think the talk will help a lot of users to report better bugs in the future, and maybe even George will sort out his DMA access now.
My talk was really rather scary for me and I was quite nervous (and unprepared!); I think it went fairly well though. The Zsh demo at the end seemed to get a few oohs and aahs.
Harry Moyes, a guest speaker from manchesterwireless.net, gave a talk on the process of setting up a charity in the UK, and the details thereof.
Also thanks to Gareth Bult for his talk on Flash Linux. It was really informative, and it looks like a very useful and interesting Gentoo-based distribution.
Thanks to the organisers, Stuart Herbert (Stuart) and Reuben Finch (grumpydog), for putting so much time and effort into the event. I'm looking forward to next year very much :).
you can find my talk in both LaTeX and PDF on my devspace. Compilation to any format other than PDF probably won't work (you'll need app-text/tetex or similar and dev-tex/latex-beamer at least, and also I would recommend dev-tex/rubber)
- Haven't posted in a while, but...
I recently brought two new developers on board: Joe Sapp, A.K.A. nixphoeni (gdesklets) and Jory Pratt A.K.A. anarchy (qmail/vpopmail). Both seem to be settling in well.
I've bumped
mail-mta/msmtpto 1.4.0. I think I'm the luckiest maintainer in the world with the package's upstream, a chap called Martin Lambers, who:- Autotools his packages properly
- Announces releases on sourceforge and freshmeat in particular so I can track them easily
- Uses Gentoo
- Is active on the bugzilla
- Is a nice guy and easily approachable over email
- Writes good software (features, portability, good code, etc.)
It makes things very easy for me, and takes a lot of the nasty bits out of maintaing packages. I've gotten Markus Rothe (corsair), who is a PPC64 developer, to keyword 1.4.0 ~ppc64 too. In the next release, I'm going to try and push the current version to stable on all architectures so I can purge all the horrible old ebuilds without mailwrapper support.
I've convinced Simon Stelling (blubb) to add
gtk-enginestoemul-linux-x86-gtklibs. This means that anyone using the multilibbed GTK+ applications (the latestacroread,firefox-binetc.) will not have to endure warnings about missing GTK+ theme engine modules on the command line, so long as they are using a GTK+ theme that uses an engine shipped with GNOME. Also, these programs will look a hell of a lot better.Other than what I've mentioned, I haven't really done much. I've been enjoying winding down from school this Easter holiday. Back on Monday though.
- In response to Donnie's post
In this post, Donnie mentioned the use of various spam filters and IMAP proxies.
I don't know about other people, but most of the spam I receive is in character sets that I can't even read. So, it only takes one simple procmail rule to filter them all out:
:0 * Content-Type:.*(big5|gb2312|euc-kr|ks_c_5601-1987).* /dev/null
It makes sense to put this sort of thing before your spam filters, as it will use nowhere near the resources.
- My nomination for the Gentoo Council
Elfyn McBratney, beu, (by the way, good work on the marriage!) very kindly nominated me for the Gentoo council. I'm happy to accept this nomination.
Well, usual rubbish as far as reasoning goes: I feel I'd be able to communicate well between projects and developers, and I think that... well, I'd enjoy the job. There's not much more to it than that.
Good luck to the other candidates.
- Re: Flashy Desktop
Spider, I would recommend media-sound/synaesthesia for audio visualisation -- presuming you're using x86. It's not at all portable.
As for the desktop side of things, one man's flashy desktop isn't necessarily anothers. I'd say stick with stock gnome as far as possible. XComposite drop shadows always look good with it.
Misc
XML Feeds
What is RSS?
Who's Online?
- Guest Users: 46
powered by
