Diego Pettenò

Okay, tonight was still a bit off after the week working on the data entry job, and all the holidays and stuff (almost all of my best friends are now happily on mountains, near Santa Croce lake, and I'm still here in the hole of the world near the city). I spent the evening and night watching About A Boy, and then I've thought it was the case to clean up a bit my site, moving a few more projects to the "old stuff" page, as I stopped caring about most of it.
Tomorrow I'll spend more time for house cleanings, and then I'll see to restart working as a full-time dev until a new job comes.

Oh I still have to learn docbook, but it seems like docbook2man requires the use of (old) sgml format, if someone knows if that's right or not, can please confirm it on comments? That would be great :)

I'm also tempted to start learning Ruby on Rails and looking for AJAX programming, it might be a good alternative solution to propose in spite of Access/VBA data entry software, it would allow to use a more powerful database like PostgreSQL while being user friendly, not counting the ability to depend not on a single operating system.
Yeah I'm probably not going to convince the people I worked for to move to such a framework, but it might be an interesting test and something that can actually show off well on my curriculum that mostly sucks right now.

It's a time when things breaks suddenly: today my old headphones broke, sigh, so I had to buy a new pair. This time it's a Sennheiser HD 515... for sure a lot different from the old pair.
As I was at the shop, I also bought the cable to connect my VCR to the new amplifier, and I also took the time to look for a DVD to give my mother as Christmas gift (knowing that she likes Richard Gere I choose a "special price" Shall We Dance DVD... actually, two DVD, as one of the two is a Windows Media HD-DVD... but well we don't have anything to play it on... it was at the same price of the single DVD version so I took it anyway), and finally as I was there, I also found a special offer on About a Boy DVD... I liked that film from the first time I've seen it, and I've seen it two or three times, but I still have to see it in English, that's the main reason why I buy DVDs.

In the last days I've also spent some more time, during the evening, watching the extra DVDs for The Lord of the Rings that I always left behind me. With them, and the new DVDs, my ~/.dvdcss directory counts 64 keys.

On a Gentoo side of the things, I've completed my job today at 15, so tomorrow I'll have time to reconnect the gfbsd box, and finally start tacking down the remaining problems. Tonight I'll take a look to the rest of the bugs, and see to handle them.

Oh I started trying out kdetv as an everyday TV viewer, as it's simpler to command with kdelirc, but it does not really seem to work that well with tuning, I'll have to import tvtime's stations.

A simple and quick update, as I've just took a forced break from the data entry job (as it would have been too good as a job, I have to do it with a faulty box, the keyboard stop responding after a while, and I have to reboot... or to shut it down for a while).

First of all, a little of self-advertising ;) Linux.com published another article of mine: Using Gnulib to improve software portability. I want to thanks Chris for proofreading and fixing it up before submission to Linux.com editors :)

I'd also like to say to the advertisers of an italian newspaper, that making radio stations air that the _first_ DVD of a series is available when it was published two weeks before and it's no more available, it's just a stupid way to gain readers..
I'm referring to the DVDs of Italian actor Marco Paolini (Italian Wikipedia), in particular to his show about the Vajont disaster, one of the most touching acting I've ever seen.
But now knowing that there are DVDs available for his shows, I'll try to find them somewhere.

Now I'm going to clean up a few things, and then I'll restart working, tomorrow is the last day, and I have still about 200 records to entry in the database.

It's time for me to start learning something new, and this time is the turn of DocBook. While I'm still going on with the job that's taking up my time, I need to relax my mind with something else between the breaks.
Today, I fixed the tree to use bindnow-flags function instead of -Wl,-z,now flags, removing a couple of userland_Darwin checks, but this takes longer and while useful it's not productive for myself :)

Why I want to learn docbook? Because unieject needs some user documentation, and the mxml2man program seems not to be what I really needed, although I hacked it quite a bit, I didn't get a hold of it enough. So I'm thinking of just writing everything in docbook, and ship it with prebuilt manpages.

Anyway, now the main parties already gone, tomorrow I need to end cleaning up my room and insert at least 150 records. At 30 records/hour, I need quite a bit of hours spent on that, I'll put the Dream Teather DVD as background music, most probably.

Okay, just for the sake of telling everybody that, I've imported the code I started for atmosphere2 (rewrite of ATMOSphere as Framework instead of a all-in-one program) in KDE's SVN repository. It can be found in /trunk/playground/network/atmosphere .
I hope people can take a look to it and maybe help with it.

This follows my plans of making public all the sources I was working on locally, so that if I'm going away someone else can take care of my old projects :)

Now I just need to write maintainers' guides for xawdecode, adding notes about zvbi and alevt that I'm going to take care, too.

Okay, tonight trying to clean my backlog I'm also blogging quite a bit.

I want to ask something to all the SVN guru out there ;)

I have some packages in KDE's svn, KNetLoad in extragear/utils and KDigest in playground/utils. The way KDE's svn is handled, I need a checkout of the files inside extragear/utils and extragear/utils/admin (the same for playground) plus the whole subtree for my package; I don't really care about the rest of the subtrees.

There's a way to tell svn to locally ignore all the subdirectories of a checkout a part the ones I tell it not to ignore? :)

I really hope there's such a feature or updating those trees would be reaaally hard... and I was considering importing into playground a few more projects to avoid stucking them locally...

Well if somebody has a solution, leave a comment :) thanks!

Okay, this post is just a recruitement call that I'm trying to launch.
Gentoo/FreeBSD needs more devs, the effort is, right now, most handled by me alone, and I cannot go on with this alone at this point. Stephen is the usual slacker ;) No, OK, he's probably having some reasons why he doesn't work on this full time anymore. Aaron, too. Michael cleaned up the current handbook, and that's a big effort.

What I'm looking for now is workforce wanting to join the project to improve it on some different points:

- somebody that knows binutils, and wants to help with porting to binutils 2.16;
- somebody that wants to coordinate effort to port GNOME to work on FreeBSD out of the box for next release... this is something that would require coordination not only with upstream, but also with FreeBSD people, if they are wanting to do this, but I doubt it (and that makes me feel really really hated :( );
- somebody wanting to split out the token ring support from dhcpcd (that's linux specific) so that it works on FreeBSD;
- somebody that wants to finish my work on baselayout porting, fixing the last issues with networking and making the current baselayout stuff to work on FreeBSD; most of the work is already done, you just need to complete networking and trying to work with UberLord and Azarah to merge them in mainline baselayout.

Already Gentoo devs and non-Gentoo devs are wellcome to speak on gentoo-bsd mailing list (or on IRC, but I'm currently not there because of the new job).

Really, the efforts required are becoming more "mundane" with time passing, so it's not a big problem. But if I'm going to take more jobs in the future, the project is probably going to stand still while I'm working if nobody else is going to help me :)

Okay, so today I got the 625 records I have to entry in the database for work. Not exactly a light job, also if I can reach the 30 records/hour.. the main problem is that lot of data are about non-Italian people, and I find sometime difficult to read correctly the names.
As I said, the job takes my "complete attention", as I have to connect the monitor to another box to work at it (a non-networked Windows 2000 box they gave me), also if I have the iBook open for emergencies, I don't follow mails or IRC or either Jabber when I'm working on it.

I'm currently waiting to clean up my backlog, as I have a patch to GwenView that Lanius gave me the ok to commit (removes an executable stack on AMD64), and xawtv that I promised to look at (before I got the job, I joined media-tv herd to help with zvbi needed by xawdecode, and I'm thinking to move there alevt that is probably more "at home" - phosphan given it up, and it's currently media-video).

Also, as I had to send it already to two people, I put the two patches for my keyboard hack on my space, hope people can enjoy them until I can find time to find a solution to the >255 keys.

These holidays are going to be strange, from a side happy, as I'm having a strangely "lucky" real life, but a part strange because of family chaos and this work that's going to destroy my nerves, I already know. But I need the money...
I also have to clean up my room as it's _really_ a mess especially after changing amplifier yesterday.
Oh and to shut up the people bitching about holidays/christmas/hanuka/whatever... Happy Yule to all.. a bit late perhaps :P

Okay, finally yesterday I got my money for the translation work! Yai! :)
So today, I wanted to buy the only thing I promised to myself to buy with them: the new amplifier. My old Kenwoord was really falling a part, and I needed to replace it, a friend of mine suggested me to take an ONKYO, HT-R330. It's _really_ good :)
I also wanted to use the optical connection, to avoid electric noise to disturb my audio.. and it worked fine almost out of the box, I just had to low a selector in KMix to let it work fine :)

A part that, today I got a job.. a shit of a job, but always a job, I need the money and I took it also if I didn't want. I have to do some data entry for a data processing company.
It's boring, it's long, and the MS Access "program" I have to use sucks. But they pay me.

Luckily should be only a temporary job, after next week it should be already done. In the mean time, my packages won't be updated regularly. I leave them for their respective herds, but usually if I take a package for myself it's because nobody else want/can take care of it.
If this means that I'll fail to update an amaroK release the second after it's released, I want to inform the bump ricers: I bump as soon as I know _and_ I have time, when I'm working for data entry I have the monitor connected to the box I'm working with, and I don't see mails nor RSS feeds. This also means that filing bug for me to bump things is not going to work. You want things bumped the second before they're released? Well you have two methods: you start doing that by yourself, or you start paying me instead of doing the current job ;)

Seriously, tho, I might end up taking other jobs that might decrease my focus on Gentoo for a while, I really need money, and until I can find someone that's fine with paying me for working on Gentoo, the things are this way ;)

Oh well, I hope next days won't be too heavy... happy holidays all :)

I'm wondering about this. In the past I wrote about the problems between KWord and OpenOffice.org about ODT handling.
Basically, a list created with KWord does not get loaded correctly with OpenOffice 2. The issue was identified before the final release, but was postponed to 2.0.1 release.
Okay, today I found OpenOffice 2.0.1 in portage, updated and tried, almost sure it was going to work...

Well it didn't. OpenOffice.Org 2.0.1 is still unable to load KWord's ODTs correctly, although they are compliant documents and they should be loaded fine. The issue in their issuezilla was verified and closed, but the problem persists.

This makes me think of two things: a) ODT is not a true solution at the moment, because its implementations does not really conform to a single standard, lists are handled in two different ways, and OpenOffice.Org, that should be the major ODT consumer and producer, is not able to manage one of the two b) the way OpenOffice.org issues are handled is faulty, they "verify" and "close" issues before the actual release, and then they are not actually fixed, it would be simpler if people could build some experimental version and verify and close bugs by themselves after the fix has been committed, but the way OpenOffice.Org is designed, it takes too much time to build to check that for a normal user, and it has to wait for the actual release.

Not like Gentoo's way to handling bugs is perfect, sometimes I'd simply like to tell people to set verified all the bugs I resolve, so that I can close them later, but that's probably not going to work that well :|

Anyway, OpenOffice.Org still has troubles handling that particular feature, that's not exactly a "small" feature; ODT is not yet ready to be the perfect interexchange format for every office suite.
And the ironic thing is that if Microsoft would have supported ODT, maybe they could have found a way to make them incompatible with everything else like they did with HTML.. are we going to use their format in the future, as actual working interexchange format, similarly how we currently use Samba to share data between Linux, FreeBSD and other L/OSS operating systems?

Today, I started getting annoyed by the continue wall output of apcupsd while the electric saw was being used. Unfortunately, the power cabling in my house makes the voltage go up and down when something that has big power consumption is started, so all my PTYs were full of apcupsd's warnings about "power faults" and returns. The same goes for my mail inbox.

So I disabled WALL warnings and started thinking about writing a KDE client o apcupsd. It shouldn't be difficult, no? I just need to write an user interface to show the data produced by apcupsd.

The only problem is that it does not provide any kind of library to access those data, there's a TCP protocol but it's not documented, the various utilities used by apcupsd uses an internal library to communicate with the daemon, and so there's no way to actually write anything without reverse engineering the protocol.

Okay, I think I have an interesting work in front of me :)

Okay I wrote today of the take over of faad2.
Just a little update as errata. Foser just contacted me on IRC, seems like we got stuck in a misunderstanding: he actually replied to my mail telling me to go on with the change when I asked him, but the mail was lost in the middle of an SMTP transfer (as he didn't maintain the CCs, we had a single point of failure).
I'm sorry for the misunderstanding and I want to excuse with him for the previous blog entry.

And for the others: remember to leave the CCs when you reply, less probability that the mails gets lost :)

Update: comments disabled as of 2006/03/22, too much spam here.

Okay, I was already using Modular Xorg in my iBook, and experimenting with it it seemed quite interestding and working fine for most of my uses.
Today, I wanted to do the hard jump and go to modular Xorg on my main system.
I tried to cleanup all the packages I needed for modular xorg (actually, just tvtime), and then moved on removing xorg-x11. I had to remove also a few packages that are still unported: opera and netscape-flash (waiting for them to be ported by their maintainers, I don't really _feel_ the need for them right now), alevt (that is something I have to fix ASAP!), zvbi and XdTv (the latter is already safe, I'm its maintainer, but the first isn't, yet, but I can build xdtv without zvbi for the time being), and x11vnc (I'm not sure I'm going to need it, I have krfb).

To be honest, I also needed to hack the evdev driver to behave as I want it to, remember my Logitech keyboard? Luckily, the code changed just a bit and I was able to deal with it.

Anyway, now I'm running modular X and it seems to work quite well, also quicker than before, I think. It might be related to the fact that it shouldn't be stripping cflags now. And I'm currently using also a -g3 build with splitdebug enabled ;) (useful if it wants to crash, by the way).

I've enabled composite for the eyecandy effect, and the current version seems more stable than the last one I tried.

It's also interesting to look at the output of analyze-x86.pl script (I always forget who gave me that, morfic, iirc, but I'm not sure):


flame@enterprise ~/devel/hacking/tmp $ ~/devel/analyse-x86.pl usr/bin/Xorg
Checking vendor_id string...AuthenticAMD 64
Disassembling usr/bin/Xorg, please wait...
i486: 13 i586: 0 ppro: 0 mmx: 2636 3dnow: 0 ext3dnow: 0 sse: 946 sse2: 3550 sse3: 0
usr/bin/Xorg will run on AMD Athlon64 or higher processor.

flame@enterprise ~/devel/hacking/tmp $ sudo ~/devel/analyse-x86.pl /usr/bin/Xorg
Checking vendor_id string...AuthenticAMD 64
Disassembling /usr/bin/Xorg, please wait...
i486: 13 i586: 0 ppro: 0 mmx: 1835 3dnow: 0 ext3dnow: 0 sse: 1036 sse2: 3662 sse3: 0
/usr/bin/Xorg will run on AMD Athlon64 or higher processor.


the modular version tries to use more the sse and sse2 extensions, and less the mmx extension... both don't use the 3dnow extensions...

Now I also need to build a chroot for the *monolithic* version of xorg to complete the porting, as I used to have one for modular.
Oh well... time to continue working.

Okay, I did something not exactly "good" today, I've took over the package of another developer: media-libs/faad2.

This because there were bugs open for it, annoying bugs, such as the block with mpeg4ip because of the way it was being built, and foser (the maintainer) is mostly MIA.

I asked him politely if sound could take the package until he was away (as he was listed as "temporary maintainer", and there was no herd to back him up), but I got no answer in two weeks (while he seems to appear now and then on irc), so I asked in #gentoo-dev and simply took the package for sound.

I've already closed two bugs (WONTFIX and NEEDINFO), and I'm now preparing an -r8, fixing my own bug for lrintf (freebsd) support, and finally resolving the conflict with mpeg4ip, yuppie! :P

The release is target stable as soon as possible, also because it uses a patchset tarball as usual instead of having all the patches lingering in ${FILESDIR}. As usual, the patches are stored in Gentoo's CVS.

I'm compile-testing right now, the -r8 is expected before night :)

That's the count of media-video bugs I'm looking at right now!
Because I had a good night tonight, I was so happy I continued the bugfixing spree I started today on media-video, starting cleaning up non-Gentoo-specific bugs as UPSTREAM, closing bugs waiting for replies and so on.
The result of the cleanup is that the total bugs assigned to media-video@gentoo.org is 110, 83 not counting bump, stable and testing requests; the query I use, that ignores mplayer bugs, goes down to 32 bugs! Yuuuuhuuu!

Now, also if I'm still very happy, the half litre of Coca-Cola I've drank tonight starts losing its effects, so I really need some sleep.

People, I like poetry! [ok that was way off topic]

Okay this might be of use to the ones that are currently tinkering with the split debug info for portage implemented by solar. I found this today: after rebuilding rtorrent and libtorrent with the split debug info to track an abort at end of download, I got a not better glibc invalid pointer result. The cause shown to be libsigc++ that was _not_ built with debug info and -g3, rebuilding it that way, solved the problem. The same happened with amaroK and taglib.

So, if you see strange crashes, it might be the case to rebuild the deps of the package... and if you have time, maybe an emerge -e world would be the safer solution.

Currently my /usr/lib/debug directories is 886MB, I'm going to rebuild gcc as I found that I was using nostrip'ed packages since at least two weeks, and that probably was the cause of the extra memory I seen occupied lately.

A few notes on past posts

I've committed quite a bit of de-GNU-ification today, and I also moved qt 3.3.5 to use CHOST instead of checking for kernel/libc pairs to build on non-Linux, it should work on DragonFly, OpenBSD and NetBSD now.

And about the Gentoo/Alt missing SVN, I'd like to take a point clearer: what we "lost" is anoncvs access, nobody ever wanted to non-devs having write access to it, that's why we have devs, at the end :) Also, infra (our sysadmins) are already working on the issue to find a solution, thanks guys :)

Thanks to Michael Kohl, yesterday night I finally removed the old Gentoo/Alt documentation and linked instead to the new Gentoo/Alt handbook :)
I've also updated a few bits of the documentation, starting from removing the SVN mirror that's no more available, and then updating the notes about the bindnow-flags function usage.

Today, I'm going to read it more again, and see what I can improve on it. There are a few topics that needs extensions.

As the bugs for de-GNU-ifying the tree wasn't so successful after a while, today I'm going to fix things the hard way: I'll just remove cp -a calls and similar from ebuilds, and make sure they works fine, and commit them.
Hope I won't upset anyone, but that's something I have to do to get Gentoo/Alt rolling.

For side notes about another previous post, I was able to build wine correctly now, but I needed a) to export CHOST and CBUILD to x86 values on ebuild b) to install binutils for x86 via crossdev to be able to select it with eselect binutils. After this, wine built fine. Now the problem is how portage is supposed to handle that out of the box, without requiring users to build cross-binutils.
And for who's wondering how debuginfo works here, I've 557MB of debug data right now, with just a few packages actually.. the most important being kdelibs, that allowed me to get a decent backtrace from KDigest failure; Richard Dale is currently looking at that as he can reproduce the bug. I hope it can be fixed before 3.5.1, so that I can actually release KDigest then :)
Now, to check better, I'm rebuilding ruby, qt, smoke, qtruby and korundum with debuginfo, hope that would turn out good.

It's a sad news, but carpaski's SVN mirror is now no more updated, so until a new solution is found, the SVN mirror of Gentoo/ALT overlay is stuck on an old version and unusable. For this reason, the only way to get the overlay, as stated in Gentoo/Alt documentation, is to use the daily snapshot present on the mirrors.

Just a friendly reminder: if you use the daily snapshot, get rid of the old copy of the overlay before extracting the new one, because we delete packages from time to time: in the last days I've got around removing a bit of packages from overlay, to minimize the occupied space.

On a side note, I broke my installation, again :P I patched the libc trying to get sandbox working, but it does not work as I hoped, and the result is that now I have to restore farragut with a FreeSBIE LiveCD.. I'm getting used to it :P

As I was wanting to find more about the Ruby crashes I got with Korundum and KDigest, today I tried solar's split debug for portage.

It seems to work fine, I get useful traceback finally, and the size of binaries is more or less the same. The main problem was with multilib-strict, that was triggered by 64-bit binaries debug info found in /usr/lib/debug; thanks to KingTaco that gave me the OK, I've made that library multilib-strict exempt and now it works fine.
I'm going to rebuild lighttpd with that too and try it out.

On another note, I'm having trouble with wine and eselect compiler, Mike (vapier) already looked at it and it seems the problem is with the multilib stuff... I have a theory about this, but I need to look at it more deeply.
The first problem is that the new eselect compiler does not seem to react to ABI var, but to CHOST var instead, and the exported one is still the amd64 one. I've worked around this by resetting CHOST and CBUILD in pkg_setup, but it fails linking because it does not trigger -m32 version of binutils, so I'm building true binutils for i686 and see what happens there...

Yep, I'm taking KDigest in Ruby on hold. Reason? kdebindings 3.5.0 is still too young to be used successfully. A part the problems I already reported, first with ruby 1.8.4 and then with and then with a missing marshaller (that I anyway got around by removing the file selector panel), now I'm getting a big problem with KFileTreeView, that reaches kicking up a PAX killing.

Trying to check the bug with rbkdesh, seems like I also discovered that's basically useless right now, as you can't create KDE bindings anyway, as Richard Dale stated on kde-bindings mailing list.

It's unfortunate, because Ruby it's a really good language, and QtRuby/Korundum could be really a killer application for it, as the PyQt bindings are not exactly the best package I've seen (building a 10MB C++ unit is crazy!). I had in mind other things that, using Ruby, would be simpler than coding in C++ directly, but until the bindings are safer, it's difficult to do something complex without stumbling across problems, small or big as they are...

I'll continue looking at it though, trying to track down the crash I've found with solar's debuginfo patch. I don't want to leave it alone so easily ;)

I might make this a weekly or monthly event, as who better than Flame-eyes might comments about Flames? ;) Yeah lame pun, I'm sorry for all the ones who read it :P

Anyway, I'm (obviously) referring to Linus's ideas on GNOME... well I don't really want to add much to them by itself, as I find his opinion almost right (while a bit too heated for my likeness :P).

What I want to have everybody here commenting on that read, is this entry from Tim Janik appeared on Planet GNOME (yeah I read Planet GNOME, and I usually enjoy it).
I think Tim actually hit the centre of the problem, I really suggest to all the ones flaming about Linus's declarations to read that entry, and to understand it.

First of all, a news for everybody who had problems in the past weeks with the Gentoo/Alt overlay's snapshots and the "Invalid keyword" x86-fbsd. After a downtime on lark (the CVS/SVN server), the snapshot generation gone byebye, and the latest commits was lost, these commits included the support for overlaid arch.list (thanks to bbj for the patch).
So, update the overlay from the snapshot, and re-emerge portage to see going away the portage warning.

On a parallel note, Robert Sebastian Gerus (arachnist) resumed work on Gentoo/DragonFly, that would be the fourth Gentoo/*BSD port :) What makes me wonder a bit is that 3 out of 4 ports are currently managed by polish (thunder for Gentoo/NetBSD, reb -with an open dev bug- for Gentoo/OpenBSD, and now arachnist).
The first code to support DragonFly started making his way into the tree (as usual, egetent, enewuser and enewgroup fixes), and I'm waiting for him to submit a few more patches in the next days ;)
Keep it up with the work, Robert! :)

My publisher still has to pay me, sigh :(

Just a quick note for the interested confcache users, that might find this post useful.
Today I noticed on the ibook that confcache was tracebacking after every new package; I seen a similar behavior on the modularx chroot I'm working with.

I looked at the traceback, and the reason is an invalid float parameter.. seems like Brian got tired writing it, and wrote a %f instead of %s when printing the 'f' variable ("%f blabla" % f)...

So the solution is simple until a new version of confcache gets into portage: just edit /usr/bin/confcache, at line 260, change %f to %s and save, the tracebacks when installing new libraries will go away :)

Remember I talked about a laptop to cleanup?
Well my first experiments with (k)ubuntu gone bad: the kernel used by 5.10 does not get along well with the 3com PCMCIA card I'm using; there were workarounds, but nothing worked at the end.
Mandrake installed, but got messed up with Xorg at the startup.

Then, I tried Fedora Core... at the start the CD-Rom didn't start right, it failed the test, died in anaconda install and so on.. thanks to my friend Giandomenico, I was able to get the Network installation wokring using lighttpd server on my main box and mounting in loop the ISO of the rest of the CDs. This way, Fedora installed, the PCMCIA card worked fine with 2.6.11 but it does not work with 2.6.14 after a full yum update, so I just reverted grub to start with the 2.6.11 while I work on the beast. Xorg had to be updated before it started working, but that is now solved, I was also able to get rid of the "postmark" version I was having with Mandrake. At the end Fedora Core installation was mostly fine and pretty interesting.

As the laptop is a ThinkPad using Mwave modem, I had to install tpctl to configure it, unfortunately it was a bit more tricky, the package needed was on RpmForge, and it was not installing because of a missing dep, and there were no packages to install the modules for the ThinkPad, so I had to install that by hand, too.
Also configuring mwave was not so simple: the mwavem package that is told to be the way to use it has a dead homepage, I couldn't find it on IBM's website; instead I found that the module was already present in the 2.6 kernel; unfortunately it asked a full I/O port and IRQ configuration, and that required me to reconfigure with tpctl to disable the IR port, that would be unused anyway; I just hope I was able to configure it right because I have a strange traceback on dmesg and when using Fedora's tool to configure a PPP connection I got a few "ioctl not supported" or something like that on dmesg. minicom seems to open /dev/mwave, but I can't see anything from that, not sure what the reason is, tho :| I'll give a better try to that when I'll have another phone line to hook to, mine being DSL would get killed if I connect an analog modem.

I must say, I wasn't expecting this much from Fedora.. it's quite usable at the end, also Gnome looks not so bad in that incarnation, I might reconsider those two project in the future, and maybe start suggesting them to n00bs whom I'm not going to admin ever after :)

I also have to say that working one entire day with three monitors display the most different things is not funny at all: I had the iBook updating (well it's still doing, to be honest), this system I was working at, and the laptop I was installing/configuring... I have so much of an headache right now, sigh.

Oh I'm still missing news from my publisher for the payment, last Tuesday my contact told me that all the people at the office responsible for my payment were on break because of a series of local holidays, and they would have returned Monday (today), so he would have told me the status that day... well I haven't heard of him today, and this situation is starting pissing me off :(

Okay, as today finally the situation of virtual/x11 was cleaned up, I'll restart porting for modular X... this means mainly that I'll start building my own packages, and then the media packages, on the modular X chroot, to fix their deps.

Unfortunately my chroot was quite old and I have to update it all, before continuing.

Now, Chainsaw also convinced me of start adding a few bumps for sound software to portage, at least the Qt side of them. This means I have to do two builds at a time.

I must say, I'm _really_ tired, also if it's just 1am, I usually go sleeping 2~4am... I also tried playing Baldur's Gate a bit to relax, but that hasn't helped.

The week "off" I took was indeed relaxing, but coming back to the bugs' reality is a bad experience.
The worse thing is knowing that what you do is mostly useless for anyone, or it creates problems to others. That's the feeling I get when I have to choose a path like I did for VLC, to force users to use the unicode version to avoid going away from upstream's descisions.

On another note, thanks to Kevin F. Quinn, today I rolled out another patchset for flac to fix the executable stack on x86.
I'm trying to understand better the requirements and scopes of Hardened project, after the "fight" I had with solar some weeks ago (that eventually made me take the week off I took), and I'm running an hardened kernel now; maybe I could try an hardened toolchain the future, but as I'm on GCC4, it probably has to wait for a bit more.

From the personal side, I'm still waiting the payment from my publisher for the translation, damn 60-days payments... they are also on break till Monday, so I won't have news before that.. I just hope to have my money on Wednesday as I want to buy the new amplifier, I can't watch at DVDs with this setup any longer.

And if there's something I hate are holidays... also this year, on new year's eve, I'll be at home, alone (well ok, with my parents, they don't count as friends tho), without anyone thinking of me and knowing that the rest of my friends are already prepared to go skiing...
sigh.

Okay, yesterday night I gone on a fixing spree with KDE.. actually starting from GwenView that wasn't working with hardened system (trivial fix, one file was in a big #ifdef that broke it on non-MMX-enabled systems).

I've fixed the problems with FEATURES=stricter and kde (at least kdebase) packages, as reported by Petteri Räty on bugzilla, now all the setXid commands are built with non-lazy-bindings, so should be safer.

I'm still looking at bugs to smash, and I'm trying to get a hold of an annoying bug in Kopete.

As I'm seeing more and more effort spent on xorg modular, and it's working quite well with Composite enabled on the iBook, using the ATi opensource drivers, I'd like to try it out and fix what needs to be fixed in KDE about it (the xkb bug is already fixed in released 3.5.0). I've ported my keyhack to get the logitech 9-bit keyboards working fine with evdev (until people at xorg find a good solution), but I can't really move, as it seems like us_intl layout is no more present in xorg.. I can live without on the iBook, but I really, _really_ need it on the main box. If someone using xorg-modular can tell me how I can replace us_intl with an equivalent, please leave a comment, thanks :)

My hopes to get KDigest working in the next weeks seems to have vanished; one of the functions I need to complete the creation of new digests does not work on Korundum 3.5.0 (and -r1 that is needed anyway for ruby 1.8.4); thanks to Richard Dale, there's a patch to implement the missing marshaller; I'll use it locally to work on KDigest, so it won't stop me, but I have to wait for Korundum 3.5.1 to get the function available on downstream distributions.

Oh well, thanks again to Caleb and Richard for the quick answers and solutions :) I hope other projects using Korudum will appear in the future, so that if other problems are present, they can be fixed soon :)

Remember if someone wants to help with KDigest, just drop me an email, it can be found on KDE's svn inside /trunk/playground/utils/kdigest if you have KDE SVN account; it uses the usual translation and documentation modules for KDE, also.

Not my laptop this time, I found a little job to get a few euros, an old used laptop to cleanup, to make it usable by its new owner without problems and mess ups.
A part a bit of problems as the CMOS battery gone away (that's why I like having a multimeter at home, it's useful to check the status of batteries), now I'm wondering what I should install on it.

The person going to use this is not a geek, and needs something simple; I'm probably not going to have to do service for him after this time, so I don't have the need to put there something that would be simpler to _me_ to administer (that's what I do to people that I have to support later, so in that case would have been gentoo the choice), so it's a bit of an experiment.

Currently I'm thinking between Ubuntu, Mandriva and Kubuntu. The first is for sure one of the most impressive projects in the last years for newbies and people that are not interested in become techies, the second was my own starting point (with Mandrake 7.1) and use KDE that's a pro for me (as I know it way better than gnome), and Kubuntu that would join the pro of the two, hopefully.
I'm downloading the live of KUbuntu now, I should have an Ubuntu live but I think it was for PPC, so I'll re-download it, while of Mandriva I have a CD-Set from a magazine.

Oh well, it would be an interesting work to do, I just hope that the internal modem works fine or I'll have to tell the person to buy a PCMCIA card, and that might mean that it won't take the laptop at all and look for something better (I hope to be paid anyway).

Well going to torrent now, also if lately I saw rtorrent failing from time to time, seems to be related to GCC 4, I have to look for a way to debug that better.

Ok after the second part, another update on the KDigest front.

I tried opening in the past days a project on BerliOS to host KDigest to allow other people to use it, but the registration procedure does not work with Konqueror. Then I remembered that exists the playground module in KDE's SVN, so I put Kdigest there; this way it's also possible translate KDigest and add documentation to it.

I'll update my site later and then user DevCounter to look for graphics for the icons and translators.

I hope it can become something better in the future, as I find it an interesting project, and might help the people who likes to do things in GUI instead of coming down to console for some tasks.

As I already blogged, I'm rewriting my KDigest application using Ruby and Korundum bindings for KDE.
I must say, the work is simpler than what I was expecting, the calls seems also more logical, and after a couple of problems now I'm able to do quite everything the old KDigest did.
Unfortunately the creation of digests is getting difficult, as one of the methods I was using it's not available on Korundum, but I found an interesting way that might be also simpler to use from user perspective.

The icons are still the dummy ones provided by KDevelop as my graphic skills are under the definition of "low"; I'll see to open a new project at BerliOS later (not with Konqueror as a bug prevents Konqueror users to open a new project on BerliOS :( ).. if somebody wants to join, it's obviously welcome :)

Okay, tonight I wanted to take a break from Gentoo work and start something that would be pleasant and useful to learn something new, so I tried converting KDigest in Ruby. Why in Ruby? Because I like it, and it's simpler to debug, and as the standard library should already provide the functions for CRC, MD5 and SHA1 I need, it's probably simpler to implement than in C++.

The first problem is that I never worked with Korundum before, so I had to learn from example; the second is that the examples wasn't working... thanks to Caleb, I didn't waste all the night debugging that, when I asked him, he told me to try with Ruby 1.8.3, and now I'm actually using that, as 1.8.4_pre1 breaks QtRuby, sigh.

Anyway, in less than 3 hours I converted the GUI initialization functions from C++ to Ruby!
Tomorrow I'll start with the true code, I hope I'll be able to recode most of it leaving the same base architecture, then I'll import everything on an SVN locally, and see to restart the work on this application that might be of help to someone, I hope :)

Oh by the way, I _really_ like the laser printers that can print without too much noise on 2am without waking up my parents :P

I released a new patchset for vlc today, and now another package require a complete patchset. It's currently having a lot of automagic dependencies, that are not fine for Gentoo, and I'm going to try make them useflags so that people can get rid of the things they don't want.
Luckily, the upstream seems alive so I can actually ask them to merge them so that next version does not require a complete patchset.

I have to say, also when I try to keep on break, there are currently too many packages requiring love, and I can't really stop patching for more than 24 hours. Tomorrow night I'll be out of home as I'll have a sort of party with my friends for my birthday (that was last monday), so I won't be at home, but I think tomorrow afternoon I'll be _again_ patching something.

Sometimes I wonder if this would ever help me finding a decent work. I don't think there are lot of workplaces where people are asked to apply patches to software they didn't wrote in the first place :P
I need to find a work, as soon as possible, I still have to complete paying the iBook and it requires maintenance, and I would really make use of more machines to help me with testing :|
Not counting that I still have to be paid for the translation I made, and counting on that money to buy a new amplifier, I was thinking of a Onkyo HT-S580, after a suggestion of a friend of mine, but anyway it will be available only after the middle of the month in the near shop :(

I already talked about the need I feel of a more complete documentation for Gentoo's packages, not users ide but more maintainer side. I'm also trying to look at what other distributions do about packages, but I failed finding proper documentation about them. They keep changelog, but finding a description for the applied patches does not seem to be available for most of them.

Instead, the documentation for package maintainers can really help maintainers of other distributions, too, by talking about the main quirks of a package, as there might be lots of things that upstream consider in a way but needs to be done in other ways to have a working package.

The only thing I see that have some information is Debian Package Tracking System, that's anyway not expressive enough for what I'm looking for.

We share lots of patches with other distributions, but the patches are often dug up when we have a big problem and we don't know what to do; having a way to comment a document the package management would help sharing the resolutions for common problems, and would waste less time.

Anyway, I've added more documentation on sound project, but I still have no clue where to put documentation for bsdtar, rtorrent and netatalk. If I can't find anything, I would probably put them on my website, but that's not handy in the same way as the project pages are :)

Avahi is another implementation for Apple's Bonjour protocol, that this time seems to be cleanroomed using glib, dbus and having bindings for Python and Qt. Avahi is also a GPL product, that makes it more appealing for GNU lovers (I usually prefer looking at the technical side of matters, instead of licensing). One of the strong points of Avahi is that has libraries to be compatible with howl and mdnsresponder, the two other implementations that are often used to enable software to use Bonjour.
Why this is a strong point? Because it allows to run a single responder instead of having to choose between mDNSResponder and mDNSResponderPosix.

To make use of this feature, tho, the programs linking to howl or mdnsresponder needs to be changed to depend on avahi and to link to that instead of the original libraries; the change is not exactly trivial.
Today, after a quick mail exchange with Sven, I added to mt-daapd the support for using avahi instead of howl, by adding an extra avahi flag conditioned to howl flag. It seems to work fine from the base view, while I can't add two mt-daapd sessions to avahi for some reason. I've wrote mt-daap's upstream hoping for it be a known mt-daap problem instead of an avahi one that might be more difficult to handle.

For now, I don't have anything else that uses howl on my system, so I've just removed it, I know there are a few programs using mdnsresponder, and kdelibs is the first one, but I want to wait a bit before trying to try getting that working.

I also have to try the patch that Giandomenico provided on vlc-devel for avahi 0.6 support in VLC 0.8.4, so that I can add an avahi useflag to that, too.