Diego Pettenò

Okay everyone using ~arch would probably have noted that there's now a new version of alsa for everyone, 1.0.10_rc2.
Why did I unmask an rc? Well the reason is actually more simple than what you're thinking: alsa 1.0.9 does not build with kernel 2.6.14. That version actually ships by itself alsa 1.0.10_rc1, and between that and alsa-driver-1.0.10_rc2, I really suggest the latter. Also, if you have an Audigy 2 ZS, you probably want to use that because it's patched with upstream patch for its DMA support.

Now I'm working on kamix to close bug #109335. Seems like there's a bit of patching needed.

Oh and by the way, today I refreshed a bit my website. I should probably change the layout into a three colums, tho, as the links starts being a bit cluttered.

Okay, today I officially removed my devaway and re-started working (as I ever stopped, eh? :P).
As a friend of mine showed me a package that was broken (tse3) I started looking at sound bugs relatives to that, noteedit and other things. At the end of the day, for sound is a quite interesting cleanup.

First of all finally noteedit can be merged fine (tse3 was broken since quite a bit before, now is fixed), and there's an updated version in p.mask waiting for testing.

Then I package.masked and sent a last rites for pmidi and daapd. Reason of this is that the first is replaced by aplaymidi in media-sound/alsa-utils, while daapd seems to have troubles with latest iTunes6, and being dead upstream, mt-daapd is suggested instead. This allowed me to close another two bugs.

Then I looked again for dupes and, as xmms was still interfering with my Sound query... I just reassigned every XMMS-related bug directly to MetalGOD :) Sorry dear :P

Now the bug count for sound is down to 110 or so. I think I have a fix for another one, but I need a testmonkey with KDE compiled with -arts and no arts in the system at all... if I can't find anyone I should probably start creating a new chroot.

Oh on another note, I talked about the SIP client I found for KDE, twinkle.. unfortunately it seems to stuck in a deadlock while closing a conversation. I'm trying to get a hold of the problem with Michel, the author. For now I'm still without client :P

Seems strange, but while I'm on AMD64 team, I don't really mark many things.. I used to have a stable chroot to mark stable things, but I haven't updated it in eons and then I just trashed it.
I sometimes marked things ~amd64 if they were missing, but it's rare.. well if everything goes like I think, in the next days I'm going to mark a couple of packages :P

The first is app-mobilephone/moto4lin, that builds and runs fine on AMD64... I have to see if it works with my new phone, tho (the Motorola V180 I talked about)... to do that I have to buy the USB cable, luckily it seems to be a standard USB mini-A connection.

Then, I was looking for a SIP client for KDE/QT, as kphone does not seem to work with the VoIP provider I just registered with... Twinkle seems to be what I was looking for, but it requires to be patched for AMD64 compliancy, and also its dependency net-libs/ccrtp needs to. Unfortunately to submit the patch to ccrtp I had to registered with Savannah.. and yet another registration, sob.

Talking about the VoIP provider I registered with... well it does seem to provide a decent service, so I can't blame them about that, but in the "Who we are" page, they really seems to be lamers:

"When the project shaped, a storm rose above the Eutelia's premises and the thunderbolts struck the iron tower that dominate them; from here the Skypho name as short for Sky Phone."

a part the "poetry" of all the story, that I don't trust neither with a nail of my fingers, for a project born in 2005, with "skypho" as name, the only reason for the naming is the similarity with the way more known Skype service.
Geeeee, why they need to be so lame while faking the reasons of their name? Better avoiding telling that, than having this lame impressions to people who know other services (because I'm sure many Italian users that does not know Skype will fall for such a beautiful story..).

Oh well... and for the KDE category I added to this post, Halcy0n confirmed that an ICE I was getting building 2.6.14 kernel with GCC 4.0.2-r1 is due to the famous visibility patch backported from Fedora. Now remember me, why we are still fighting with that? Because KDE upstream still thinks that there's no point in disabling visibility, also if 90% of KDE becomes unstable after that? Because Dirk is still convinced that changing the default visibility, possibly breaking KDE in subtle way is good?
Sob, I just hope that people don't start blaming "KDE is unstable" just because of that visibility stuff that does not and won't ever work.

I still result in devaway. I'm not really totally away, as I'm still on IRC, on mail, I take a look to bugmails and try to fix the things I know how to fix. It's more a way to make sure that nobody expect me to be responsive "on the fly" as usual. I hope to remove the devaway soon, but until I take the driving license, I have to spend more time in driving lessons than anything else.

I took anyway the time to clean up a couple of (Italian) articles I had on my site, while I was writing the presentations for Italian Linux Day (that will take place at end of November). Specifically, I restructured the one that talk about using DynDNS to simplify the access to boxes inside and outside a network, that used to explain how to set up DJBDNS, to explain instead how to set up dnsmasq, as that seems to be more suited for the task, and quite performant IMHO.
I also updated the article talking about setting up a local mail server using Fetchmail, maildrop and Courier IMAP, to use instead Dovecot as IMAP server, reflecting what I've changed locally.

This is a quite interesting task, as I done this changes after a while I originally wrote those articles, and after changing the software I'm using myself. I think this is a good example of how the time changes the requisites and the solution to the same identical problem. It's also one of the reasons why I don't like people that fossilize over a single solution without looking elsewhere.

I should probably try to translate them in English, but I need help with them after the translation (to remove the big screwups with grammar) and spb is slackering also with the last one I asked him to read ;P

Oh well I hope at least the Italian people can read them and find them useful, somehow..

Ok for who followed the two episodes of my problem with the old mobile phone, I want to say that I have now a new mobile :)

I found a good, although not "latest model" Motorola, V180, at 100 euro.. not so bad after all.
Also, it's one of the last phones that are actually phones, and not everything-but-espresso.

After this, I also removed the V300 from my funny (or not-so-funny) wish list, as this phone is enough for what I do :)
Also, it's quite elegant, light and comfortable to use :P

In the bad situation of having a mobile die, I was able to get something good for the minimum price :P

Seems like current media-sound/daapd package has some problems with iTunes 6. Unfortunately, it's also unmaintained upstream.

I'm now wondering if I should p.mask daapd and give a last rites to it, in favour of media-sound/mt-daapd that seems to work fine with latest iTunes (and it's actively maintained).

This also comes when I've moved from net-misc/ntp to net-misc/openntpd (that finally seems to work fine on Linux, aye!). I love when there are alternatives :)

On a personal note, yesterday I said my mobile was dying, well today it died :|
Now I'm without mobile (and in Italy is not exactly a good thing, as I'm usually in touch with my friends through it). I had a spare Nokia 5110, but it does not work with my provider (as it requires dual-band at least) and it would reject my simcard (32k).
It's time to look for a new mobile.. the problem is that a Motorola V300 still costs 150 euro... yieks!

I'm currently on devaway.. for a couple of days more.. I had to take a break to take again the control of myself. I was able to fix myself about this, but I think I'll enjoy a couple of days of break again :)
I'm not actually away anyway, I'm here, I'm just on reduced workload. I'm looking at bugs, and I try to fix the ones I know... I don't really look three times at them as I use to do.

I'm also taking driving lessons as I have to take my driving license before December, if I want to move out of this place out of the known world :P (and possibly find a job)

But what is the title of this entry? Well my mobile phone's battery is dying... again.. it lasts about an year before I have to replace it... and it's the third, it starts getting expensive.
Also my audio amplifier is broken and I almost completely lost the left channel.
And when I'll have my driving license, I have to find a car...

I need to find a job ;_;

Seems like this night I wasn't able to sleep. The reason is probably that I'm a bit confused about myself... I'm thinking of things that I shouldn't, I'm thinking of a person I shouldn't...
So, after two (2) hours of sleep, I took my iBook and drained his battery to the end hacking around.

First of all, I've tried cleaning up my Gmail account.. I receive there many aliases, not only Gentoo's but also a few other addresses I used to use (like Sourceforge's and Berlios's), this means I received and archived also bug mails and similar stuff, that is not exactly things I like to have archived for long times. So I looked better and found that you can set up filters to redirect mail to other addresses and move it to trash.. that is *really* interesting, so now I can use Gmail's spam filter with all my Gentoo alias, archive discussions done on those aliases, avoid receiving dupe mails I send to the aliases (:P) but *not* archiving the bugmails. Yuppie! :)

Also, I re-checked the ebuilds that should have been fixed for Gentoo/ALT, as there were a couple of extra enewuser calls that used /bin/false... I should ask if I can add a QA Warning when calling it with that specific shell....

I also finished work on media-sound/daapd, as there was a remaining bug.. -r2 should be fixed. I also thought of adding the symlink support to daapd but... I'm not sure, the init.d is quite complex, and daapd itself seems to me like a bit abandoned.. mt-daapd works better for me, and has a less dumb build system (static makefile vs autotools).

I merged netatalk into main portage, as I tried the fix over and over and seems to work fine, not even requiring the m4 package anymore. It still runs autotools, tho.

Hopefully if Michael finishes the handbook for Gentoo/ALT, I'll add this week more documentation about porting true code from Linux-only to platform-independant...

Oh and yesterday I restarted taking driving lessons so that I can take my driving license for end of November.. I really hope so!

This morning is dedicated to Sound Team :)

The first big news is that this week's GWN reports Sound Team as first in the closed bug ranking :) Thanks to Andrej "Ticho" Kacian (this time I got it right the first try, I hope) and Luis "MetalGOD" Medinas for joining the team and taking care of MPD and XMMS :)

Then, a new amaroK was released. I wish to thank the amaroK team for their pre-announcement to packagers, that is a really useful feature :) Unfortunately, when it was sent I was sleeping, so I'm bumping amaroK now (so yeah, I know it's released, I'm going to commit it soon, don't open bugs about it, thanks :) ).

Oh by the way, the new amaroK solves the problem with non-ASCII collections, that is a good news for everyone using UTF-8 or other non-ASCII codepages (so basically quite everyone that lives in a non-English-speaking country :P).

Committing amaroK now...

Yesterday night I had a sort of party with friends, so I'm currently a bit blurry-minded... not like I drank too much, as I'm abstemious, but just a bit tired. Funny night tho, so I'm also quite happy.
Unfortunately, I basically knew noone there, but that was not so much of a problem after a while...

But this entry has a title, that refers more to yesterday's afternoon, waiting for my friends to take me up .. :P

In the last days I'm reading Irregular Webcomic!... and I'm finding it the craziest, geekiest, best webcoming I ever seen :P
I still think the masterpiece is this one (note, you should be at least a Star Trek geek and probably an Harry Potter fan or at least know the characters' names to understand this one).

Ok now the question.. what is *your* favorite webcomic? :)

*Starts tinkering trying to create a Insanely Overpowered Fireball*

Ok the new Release Candidate for XOrg modular was released, and Donnie added it to the tree, so yesterday I wanted to try it on Gentoo FreeBSD (no slash, why? I'm trying to see how it looks like without, following Ulrich suggestions...).
I removed it from the main installation, and I planned to play with it on a chroot, so I had to predispose the chroot that I haven't before. After that, started merging all the deps.. stopped as usual with mesa, as I had to patch it to work on FreeBSD (as usual it's on Gentoo/ALT overlay), and then started xorg-server.

Took quite a bit more than usual, the bsd step finally seemed to work, but then it stopped with undefined reference. Yesterday I was too tired to investigate, and today after touching mt-daapd (that is now in tree with multi-service support as mt-daapd-0.2.3-r1) I gone taking up my new glasses (the new green lens seems to be cool, but I still have to be used to them, so they are giving me a bit of an headache, I'll probably take the night off watching TV with the other pair). Now I wanted to look deeper at it.

The problem is ix86PciInit funcion undefined in one of the internal libraries, that gets an undefined symbol while linking the final executable. So I go searching for that function, that is in bus/ix86pci.c file.. .that is not compiled... I start looking at the reason and I found out that the file is built only when there's no better OS support for it, so it's disabled for Linux, FreeBSD, OpenBSD, NetBSD... fine then, but why it's looking for it?

Looking again, I find that the init function for PCI device is chosen in a list of functions depending on the hardware platform and the operating system.. FreeBSD uses its own version for IA64, PPC, Alpha (iirc) but not for x86, just Linux has its own version.. but that collides with the configure, that disable that file for Linux, FreeBSD, NetBSD, OpenBSD.... there are at least three platform/os combinations that won't work.

Fine then, the problem is -again- not in portage, but on XOrg itself... I submitted it to upstream as usual, waiting for news.

So today I worked on a new stage for Gentoo/FreeBSD to fix the ncurses problem.. it's probably hitting the mirrors right now. A part that, I started another problem.

I usually leave separated the CDs that I rip to play with amaroK in different directories on different disks, this means that mt-daapd can't index both of them. I had to turn to mt-daapd because daapd stopped working in stable way with new changes... the problem is probably due to CFLAGS as the problem are not present for many people. I'll try to investigate more when I have time or find enough people to have a trend, in the mean time mt-daapd is working.

This daemon has a problem, tho, as it requires files to be in the same filesystem, so it does not work with different directories on different disks. What could I do? Simple, I implemented Azarah's and UberLord's approach, and started working on the symlinked init.d scripts.
Now I have a beautiful /etc/mt-daapd.d/ directory where I can put <name>.conf files used by /etc/init.d/mt-daapd.<name> init.d (if there's no name, it uses /etc/mt-daapd.conf as usual).

Unfortunately, there are a few issues for example it uses always /var/run/mt-daapd.pid file for storing the process id, so I have to hack it to allow me to change the file.

When this will be solved, I'll prepare a 0.2.3-r1 for everyone's use :)

I can't say I love autotools... I don't. But I much prefer autotools to broken and messy manually-created buildsystems which you must ask them *where* you're building, or what you want to use. libtool is an hell, many times broken, leaves .la around making your lib directories a mess, but it works.
Find me another build system that have the same features autotools has, like autotest for headers or for libraries, user-friendly parameters, support for prefix, CHOST handling and crosscompile help, and I'll start using that.

In the mean time ncurses gave me an headache, because today after upgrading to 5.5 version, neither /bin/ls worked anymore on Gentoo/FreeBSD. The reason? ncurses uses the full version as soname... this means that 5.4 and 5.5 have different soname, and breaks. I worked that around creating a symlink libncurses.so.5.4 and now I'm working to get ncurses to use a decent soname... if that works fine I'll commit in Gentoo/ALT overlay and then ask vapier to commit it.

This also means that I have to prepare a new stage to avoid new users to encounter this problem... and this is a stage to redo basically from scratch as most of it needs to be reinstalled... sob...
I'll take the chance to fix a few more things that needed a bit of attention, too.

Okay, today I felt a bit tired, probably because of all the flames, the problems, the solutions and so on. And when one is tired, the most effective thing is to relax hacking a bit on something new. My usual "something new" is atmosphere project, for network monitoring in KDE, but today I wanted to spend more time on something really really new that could help. What it is? UPnP

Universal Plug'n'Play, whilst being probably an ass for Business and "need-to-be-secure" LANs, it's usually useful to manage home routers. Also, in a controller environment, having the ability of adding/removing ports forwarding on the fly is probably simpler than having to predispose all the forwarding from the start.

For Linux there is an official SDK by Intel... but I had to patch it quite a bit to have it working on FreeBSD (ports carries an old version of it).
Then I started reading doc and hacking until I was able to enable the forwarding of a port on the router... and this was the most difficult part of all it.

What I want to do now? Well my first thought was to add UPnP support to rTorrent, so that I can drop the port range from its configuration, but I'm currently on break as I have a bit of an headache right now.
Another thing I'm planning for is creating a libkupnp project to create a simple UPnP library for KDE applications, to use in ATMOSphere, but to use also on Konversation, so that I don't need to specify a port range there, too.

Being able to use UPnP makes probably simpler the experience of a newbie user using Linux.. think of programs that automatically configure to work behind a NAT... this is sometimes the showstopper for new users, and the reason why some P2P freak still suggest users to take USB ADSL modems...

Another interesting thing that I'm not sure I actually want to work on, tho, it's a daemon that tells the router which port to open for SSH/IMAP/whatever, and maybe tries to register the ddns with the router itself...

Oh well, I think I'll look forward more about this in the next days, maybe with the new pair of glasses.

Probably a bad joke as a title, but considering that I slept only 6 hours this night it's the best I can find :)

Yesterday I wasn't present on FreeNode, and that is something strange... the reason of this is that I had to go to the oculist for a check up .... and I don't really like the news.

As the Planet readers know (as they see my scary face every now and then) I already wear glasses, since three years ago to be exact... now, I knew I had to replace them with something stronger as they were a bit weak being the first pair I ever had to wear... but had anyway bad news. I had to order two pair of glasses, as I need one for work, and one for drive and the rest. This is not exactly the most comfortable setup...
Anyway, my new glasses will arrive on friday, in the mean time I still can use one pair only.

A part this little problem, I'm wondering on changing router sooner or later. Mi current router is a D-Link DSL-500 Generation I, it's a Virata Helium based router with many cons and a few pros. It works quite well for "basic" support, and it has SNMP management that makes it interesting to work with. Unfortunately I cannot use the internal DNS proxy, so I have to use djbdns or dnsmasq on Enterprise (my amd64 box) to resove local names (if I try to use router's DNS proxy, after about 20 hours it crashes badly with debug access open on serial console).

What I'm looking for is a router, possibly linux based, which can be managed in some ways better than the Web interface, as that's usually uncomfortable for someone used to have terminals around :P I don't really need a complete ADSL router, probably a generic router with WAN interface is enough, as my current router can become a bridging ADSL modem, and leaving that "external" allows me to avoid having to change the entire router if I change connection type. Instead, I'd like to have something with WOL capacity... not as "server" but as "client"... just to make an example, if I shut down enterprise before going to sleep, then I wake up and want to listen to my music from the ibook (voyager), I have to go in the other room and start up the server.. WOL would probably be a good way to solve this (as voyager is connected with wireless, last I tried the Access Point filtered out the WOL packets, so I can't simply run it from here).
Anybody has a suggestion? :)

Talking about routers, does anybody know a UPnP library better than the "official" one? I'd like to tinker a bit with it to try enabling port redirection "on demand", but I don't know where to look for it :/

This night I wasn't able to sleep, probably for the pizza I ate yesterday night (22.20 is too late for that). I wake up this morning and KMail starts NOT showing me my mails.

A part the little enthusiasm for be without bugspam^Wbugmails for today, that was not good. Yesterday I had problems with maildrop, that turned out being a problem with version 2, so I reverted on 1.8, and found out that -Os breaks it. Apple's Mail wasn't showing me the mails since a few days. This was enough.

I stopped courier-imap and looked for other solutions, I heard of someone, if I remember correctly was Roy, who moved from courier-imap to dovecot, so I gave it a try... okay the configuration is a bit more complicated than courier's but... worked quite fine. Now, KMail is more responsive, the organization of the folders is a bit more human (instead of having everything as subdirectory of INBOX), and also Apple's Mail works fine. Great! :)
Now I also have a package and a service less, as I don't have to use authlib for that..

On the other hand, I worked a bit on daapd, too. I fixed the ebuild as there were a couple of bugs open, I also made sure user CXXFLAGS were respected, and made the init.d working. It worked great.. until I added more music to the archive, and then stopped working. Probably some file is not fine for it :/
Then, I tried looking at mt-daapd. It worked fine, too bad it requires all the archives to be on the same filesystem as it uses inodes to count the files. I think I'll try to handle that using Roy's method for multiple services, and then see if I'm able to get more than one mt-daapd running with different music directories.

About myself, I have not so much to say actually. Today, is exactly six months that I work as Gentoo Developer :) This is really great, I'm enjoying the development and the people are mostly lovely to work with (a part a few flaming, but as solar said, it's my fault for taking an "hot seat", I knew that when I started :) ).

Unfortunately, as my publisher is going to pay me only at end of November, and that money is probably finshed before I can actually see it, between university and driving license, I need to find a job soon... what I'm trying to look for now is if some computer magazine in Italy is looking for contributors... also a simple payment for some article can be enough...

This is linked to the fact that my wishlist, that started up as a funny way to tell people my likes, is filling with useful things that I'm looking for but I still have not enough money to buy.

I'm looking forward to find a job that does not fill my day anyway, as I want to continue working on Gentoo projects.. maybe after I take my driving license I can look for a job where I can work for Gentoo projects directly :P No I don't think so, I'm working on areas with minimum job-related links, and I'm not so good as a programmer or whatever else to be paid for that.

Oh well, we'll see... I'll let everybody know :P

Been a while since last I updated the status of this project, so here we are :)

The first important thing to say is that I was able to set up distcc + crosscc to crosscompile between my Gentoo/FreeBSD system and my Gentoo/Linux AMD 64 system (way faster).

crossdev was the saviour in this case, also if I needed to fix it a bit (thanks vapier for merging the changes). Now is possible to have a stage0 crosscompiler for FreeBSD on Linux. The problem is that freebsd-headers and freebsd-lib are not crosscompile safe, so I have to fix them before being able to have a C++ compiler (that is much needed to recompile KDE again :D).

Modular Xorg is not working yet, but it seem sto proceed well.. I hope anholt can have it working soon, and I trust his work more than the bug wrangling of another well known xorg developer (yeah, I'm still ranting about him).
I also decided to use a chroot for modular Xorg on that machine, as I have enough space there, so that I can give a try to X11 apps using xorg-x11 in the mean time.

The libtool patch that was already applied by libtoolize is now applied by libtool itself (no revbump, just re-emerge libtool to have it applied, so that more libraries get the "new" linux linking style), hopefully the testign will prove that good enough to go on upstream sooner or later.

The gentooalt-m4 idea is gone away, it won't continue, I'll change the packages to use simple patch + sed to work, I'll refine m4 files and see for sending them upstream instead. The package will be removed tomorrow from portage and all its dependant packages will be fixed in the overlay with the new method, after that, I'll see to send them to main portage maintainers.

A new stage is on mirrors, as my previous stage had a few glitches that were a bit annoying. Also, as libarchive changed soname, the old stage broke after the first libarchive update, no bsdtar, and system quite difficult to manage. For whoever had this problem, just need to symlink libarchive.so.1 to libarchive.so.0, re-emerge bsdtar, remove libarchive.so.0 symlink. The new stage does not require anything, it's built by default with libarchive.so.1.

I wish to thank again egore for its Athlon 700, that can run all the day long and the night while building, whilst my Tbird 1000 can't (too noisy), as that is allowing me to continue the work :)

Time for status updates, I really like them... ok maybe not, but they are useful.

First of all, the public scary feeling that my plan for phasing out XMMS reached its target :) Luis "MetalGOD" Medinas joined this evening the sound herd as new XMMS maintainer. Good luck Luis, and keep it up :) I think many users now can be less worried about losing their preferred player.

Also Andrej "Ticho" Kacian joined sound herd, handling mpd and related, so also this new packages familiy is actively maintained now. There are, tho, other packages that needs help and love, so if someone wants to join, it's always welcome :)

I also tried cleaning up some of the bugs for sound project today, as there were a couple that were trivial, and at least 3 referred to faac were basically the same buggy bootstrap script being used. This should be solved using eautoreconf.

On another note, I wish to thank djm on #gentoo-media, that solved a little problem I was hitting my head against tonight while trying daapd (needed to start mDNSResponder service :P).. now, using daapd and iTunes, I can *finally* listen to my music from the iBook without having to fill its hard disk! :)

Okay for this status update is enough, proceeding with the next one ,) [Planet readers will hate me so much, for showing my face every time :P]

Okay as I already wrote about this, and also Chainsaw told this quite a few times, XMMS is currently an unmaintained, obsolete, buggy piece of code. Eradicator stopped maintaining it on Portage, moving all the bugs to sound. I tried cleaning them up a bit, but there are anyway too many bugs open.

I again try to ask if someone wants to take care of xmms (alone, as sound herd doesn't seem interested in maintaining it right now), else, in 7 to 10 days I'll start the hard phase out time.

What does the phase out mean? First of all, the new ebuilds for plugins on bugzilla will be closed as WONTFIX as the other true enhancements. Then, all the plugins that have at least a bug in bugzilla will be masked and removed after 15 days.

This will continue until it will be clear that xmms itself needs to be removed entirely, in which case it will be one GTK 1.2 user less in portage.

If you're using XMMS currently you're invited to try other alternatives, such as BMP, BMPx, amaroK, Rhytmbox......

As I already blogged about the day started and continued with a quite bad average mood.. With Luca we started looking for video packages that deserves a rapid (and possibly torturing)^W^W^W death, but in the average, there was no big deals on the horizon.

This evening was instead more productive. With cryos we started looking again at a quite annoying KDE bug with kcheckpass not building on AMD64. Bug that I couldn't reproduce... and neither cryos, before tonight, at least on its laptop.

Looking at the way the target was compiled, there was no reason why it shouldn't have worked.. after spending time looking at the Makefiles "in pieces", I looked at a diff between the two and... well there was a big difference between the two: I wasn't using --enable-final, while cryos was... so what was the problem?

The problem was (and at the moment, is) within --enable-final handling of per-target CFLAGS. I've opened a bug on KDE's bugzilla to report it, hopefully this can be nailed down before release. We have a simple way to workaround this in case, but I'm not really sure if I want to do that right now, so I'll hope for the release candidate.

After founding the root cause of this nasty bug, I think the day is not completely wasted, So I'll enjoy having a working system with 2x1GB swap partitions configured, and I'll go reading something before sleeping.

On a side note, I almost fixed site, that should now mostly work on Internet Explorer, too. It might have funky fonts, because of a missing functionality inside Konqueror that masked me a little problem with my XML/XSL files, but I should have fixed it on most of the pages.

Today was a long day, but I haven't done too much :/

Woke up with a big headache, wasn't feeling well yesterday. Restarted Enterprise, just because lately it seemed bloated, then, while restarting upgrading KDE... system stuck seeming swapping. Had to force it down and start again. After a while, again. I started thinking of xine/amaroK, but that was not the problem.
Tonight after about 9 hours trying to find the problem, I try booting with an older kernel, and I look this time carefully to the boot log... "can't find swap space signature" ... oh damn!

Moral? I was running a system without swap, with KDE, GTK+, amaroK loaded, and trying to build C++ programs.... that was not good :P

On the other hand, me and lu_zero today worked on looking for some video packages known broken to remove, so some of them are going to be removed from the tree entirely soon: avifile, vcr, zphoto, drip, divx4linux, quicktime4linux. In particular, quicktime4linux 2.x seems to be a complete hell for portability, and 2.1 source tarball gone from 5 to 30Mb because it was created with .svn directories inside... see what happens when not using a build system that allows you to make dist ? :)

I'm strangely enough being a documentation freak lately.

Not only I reorganized and wrote most of the Gentoo/ALT documentation, but today I started writing some notes for maintainers about xine-lib and related packages.
The reason of all of this is to allow someone who wants to step forward to maintain something to look at those notes, and maintain them updated, so that there's no need to enquiry me for something that might be trivial.
I'll do something like that also for vlc and probably amaroK, so that if I'll be MIA or with reduced time, nothing will start screwing up.

Also, I think tonight I'll take a deeper look into writing the replacement for GLEP22, I asked on gentoo-alt, but seems like there's no one that followed the question enough to extract a good summary of that. I asked Chris White to take a look, but I'm not sure if he'll have the time.

Now this is strange for me, documentation is something I really hated to do, but lately, I'm thinking that my work is useless if I don't document every step of it for someone else who might come. This is probably due to the fact that I took the place of many people who did stuff without leaving a clue of why they did that, so I refrain from doing the same to someone else.
Also, I think I like GuideXML :)

On a similar note, I'll have also to prepare the other presentation for the Italian Linux Day... Francesco convinced me of doing a presentation out of my "Helping Free Software without being a developer" article (it's in Italian, published on my site... I know you get the data listing, I have to fix that creating an articles page and redirecting there).

Currently, the video herd is understaffed. We have too many packages and too few developers working on it. Today I cleaned up at least a few recent bugs, and there are packages mainly unmaintained.

DVD::Rip is one of them, but it's not the only one. I want to try asking here for new developers, before starting moving dvdrip and maybe other packages into maintainer-needed...

There are packages that are mostly unmaintained upstream, some of them are still needed as they are dependency of something else, others might be removed (we still have realplayer binaries hanging around that should go away sooner or later).

The same goes for sound, as there are packages there that are absolutely absurd, or things like xmms that is no more maintained upstream and we don't have anyone looking at it, mainly.

If you want to help, or to become a developer for media herds, take a look to Video or Sound bugs.
Most of them can be fixed without commit access providing patches that will be merged.

I'm trying to get something working on this side of things, also if I never read the various guide on how to set up cross distcc.
I hope that using the new eselect-compiler from eradicator it would work smoother.

I'll be back with information about how to use this as soon as I can.

By the way, I'm still working on getting some news for Gentoo/ALT. Unfortunately I suck at writing tech docs that should be a GLEP, and I have at least a GLEP to write.

I have to continue the work, but tonight I needed a break to try to clean up my mind :/

I know this is a stupid blog entry, but was just a way to say that I'm not hanging around...

Okay, probably for most people TV is already an alias for stupidity... actually, I enjoy watching it from time to time. Some TV series deserves a bit of respect, some film is quite interesting, too, and sometimes we get also good theatre shows here.

Unfortunately I lose most of the "good" shows as they get in the public service channels, and I can't receive one of them where I live. Also, I don't have a DVB-T receiver, so I can't even make use of the subtitle/other language features.

But lately, I'm getting tired of the ""new"" formats they are airing now... starting from realities that I hate (at least they stopped after five Big Brothers), going to shows were people just go begging for their 5 minutes of "glory" (and probably an entire life being laughted at from anyone with at least a bit of brain).

Then there are the new realities with "famous" people... with money rewards... to people that already have a lot of money... okay that would be good if to pay for them are just commercial stations... but public channels, that are forcefully paid by anyone in Italy who owns a TV-receiving device (television, tv card, vcr) or an aerial ....

Also, most of the commercials you see on air are stupid, clueless, idiotic and insulting for the common sense.. and we used to have commercials that inspired new cartoons!

On a different note, I'm starting to feel the need to find a job, but looking for something that would allow me to continue working at this pace on Gentoo Projects is not simple. Also, it's difficult for me to find something Gentoo-related, as my "working areas" are probably the ones with less possible commercial involvement (Gentoo/ALT, Media-Video, Sound, KDE)..
To be honest, I'm currently looking forward for an Italian magazine to write for... at least I can do something that I don't completely dislike, and help people at the same time :)

Oh well we'll see.

This happened to me quite a few times, so I think it can be useful to others, too.
If you usually use quilt, you might have experienced the common mistake of running vim on a file directly instead of using quilt edit..

It may seem obvious for most of people out there but you sometimes don't think of obvious things .. Alias vim command to quilt edit, it's a no-go if it can't find a quilt structure, it works fine when you're inside a quilt-controlled directory. If you prefer not to have always to use quilt edit, you can alias it just when you enter the quilt managed tree:

alias vim='quilt edit'

and then you are fine. :)

Ok that's enough for a stupid tip, I know.

Okay this is where we start with the problems and we need people.

Sandbox is almost working on Gentoo/FreeBSD. Unfortunately something is not completely right. Probably because we had to not use true getcwd() but the egetcwd() replacement, both findutils and coreutils configure stops at:

checking whether getcwd (NULL, 0) allocates memory for result... yes
checking whether getcwd is declared... yes
checking whether getcwd handles long file names properly...

The solution is to disable sandbox. But this is obviously just a workaround.

I can't really start working on that with my current tasks going on, so I need help for this.

If somebody is interested in fixing this up, set up a Gentoo/FreeBSD system, and then install sandbox from Gentoo/ALT overlay and finally try to run the test program (you find it in getcwd-path-max.m4 macro file in coreutils sources (inside m4/) directory.

Join #gentoo-bsd at Freenode and I'll help you if you need. If you have a patch, submit on Gentoo Bugzilla for Gentoo/BSD product (or Gentoo/ALT product, Gentoo/FreeBSD component, when we'll move to that).

Not sure if everybody knows that I were working on trying the visibility support in KDE as there is KDE's bug #109386 that makes it sound like The Best Thing After Sliced Bread!

Well tonight I reached the final stage. I hope the rest of KDE herd, but for me, it will pass a long time before we can talk about visibility support in KDE on Gentoo.
Why? Because there are too many things that needs to be changed, patched and so on.

I had gpg-agent gone crazy, I found out pinentry was not running, I tried recompiling it and what? It failed! I worked around it thinking a QT 3.3.5 change, instead it was visibility stuff, again.

KWord doesn't work. Reason? Something in QT is not being exported. It doesn't load. It doesn't even compile again.

This is a ranting entry like nothing else I did before because I'm quite tired of testing this stuff as everybody says it's clear, that there was indeed a problem before but it was with Qt, my crashes were caused by a crash in GCC, not fixed till 4.0.3 (patch backported, the crashes are there anyway), that everything is wrong because of something else.

Qt is not ready to have visibility support in the 3.x branch. GCC support is still incomplete, we have bugs. libstdc++ itself doesn't play nice with it. It's a bleeding edge feature. And you enable it by default ? Pleeeeease!.

So please don't go any more further. Visibility support in KDE on Gentoo is *not* going to be added soon. KDE 3.5 is not safe, Qt 3.3.5 is not safe, nor will be 3.3.6, as the changes are going to be non-trivial.

Said that, I hope this address everyone who wants to ask for this support now because "kde supports it".

It is not supported enough, it's too much experimental. Don't play with it if you want support from us.

Tonight my system will rebuild qt, kdelibs, arts, koffice ...

Okay everybody, I ask here for a bit of help :)

ServoFlame is my own bot on FreeNode (you can see it on some #gentoo-* channels, mainly following me, but not only). It has the interesting feature of being able to access multiple bugzillas, using the XML export interface provided by bugzilla itself.
It's not so bad, it works fine for most of the cases, but now that I'm running it on Ruby, I have problems with bugzillas that are https only.

The problem is that using ruby open() method with http/https urls fails when the certificate is self-signed, throwing an exception that closes entirely the connection.
Does anybody knows how to work around this problem and tell ruby/openssl just to accept every certificate, valid or not?

Thank you very much for this, it's something that is really bothering me right now, because it's the most important feature lost between python and ruby right now :/

[Ok it's true there are not so many https only bugzillas, but there are a few at least]

Okay as I anticipated yesterday, today I spent the day reorganizing, fixing, extending Gentoo/ALT documentation.

I started a document about starting a Gentoo/ALT port. It's very far from completeness, and it lacks many things. It also doesn't completely tell how to move from a bootstrapped system to a working stage. Tomorrow I promise I'll try to complete it as much as possible.

I also tried to work on a sort of TaskXML to document deeply some complex task like the sandbox porting and the de-GNUification of the tree, but the result I had were mostly rough. I sent them to Paul anyway, maybe they'll get accepted sooner or later, in which case, there will be no reason for actually starting documenting every step of development.

I think this is one of the few things I take as good from Software Engineering...

I said I was taking a break, but as I already did, I wasn't able to maket his a reality. Also if I employed most of the day with installing Solaris, during the evening I restarted working on Gentoo, bumping amaroK (thans Ian for sending me the mail for the prerelease tarball) and moving stuff from Gentoo/ALT overlay to main portage. Also, now that I should sleep I am planning what I have to do tomorrow!

In the past months I tried to use all my free time on Gentoo, also to avoid thinking too much. I think this is not going to work any further... I need to find a way to clean up the mess of my room and my mind.

This does not mean I'm going to leave Gentoo, should be clear :) I think this is the best thing that happened to me in the last years. I'm just saying that I need to find time for both Gentoo and my Real Life.. this won't be an issue, if I'm ablet to document all what needs to be documented before.

See? I start trying to say that I can't stop working on Gentoo... and I end up talking about Gentoo... Damn, I'm a work-a-holic :/

Okay this night starts like one when I'm not able to sleep. I think this is due to the many screwups I usually do in my real-life, but that's not the point.

Today, also if I said I took one day off, I continued working on Gentoo, at least the night. I updated the linker detection to recognize Solaris linker, and I then moved bindnow-flags to main portage. Tomorrow I'll probably also move gentooalt-m4 package, that contains two m4 macro files used to have AM_DL and AC_BINDNOW macros on configure scripts.

This also meant that I had to update the maintainers notes to tell developers how to write ebuilds that works fine for Gentoo/ALT. The new documenttation also updates the policy doc that is still work in progress. I'm now going to take note about a couple of things I want to add to the documentation tomorrow.

Grobian also update the documentation, with notes aobut keywording things for Gentoo/OSX. I wish him the best for his work as Gentoo/OSX lead, he's really doing well.

Tomorrow, I'm also planning of writing a new document explaining what things are needed to port Gentoo on a new, unsupported operating system, how to start, what to look for, the tricky parts of portage, eclass and ebuilds that needs to bhe checked.
It's a lot of documentation needed on this front, as that was always a sort of "black magic" thing.
Unfortunately most of the hidden work was always undocumented, nobody wrote anything on how Gentoo/FreeBSD was created. I also wrote just a few notes about how I wrote the ebuilds and so on. I'll try to fix this in the next days, adding more notes for Gentoo/FreeBSD development and in general Gentoo/ALT portability.

Something that I'm feeling a growing needing for is actually a Gentoo/ALT bugzilla product. Right now OSX and BSD have different products, this to allow the assignment of bugs directly to the right aliases. This works, but makes difficult to categorize the new portability bugs that involves both projects? Also what will happen when other Gentoo/ALT projects get official?
If anybody knows bugzilla enough to suggest a solution, it will be welcome. Right now the only thing I can fing of is removing the components from G/OSX and G/BSD (that are fairly small amount of splitting things up) and then adding a "General Portability" component, a "Gentoo/OSX" component, a "Gentoo/FreeBSD" component and so on, as I know the components can change the assignee. For the rest, I'm not sure if different components can use differnt versioning (that would help to select the version of FreeBSD/OSX/NetBSD/whatever that is being filed the bug for.

From a structure point of view, instead, I'd like to add an Arch Testing program to Gentoo/ALT, with ATs for OSX and *BSD. This would probably help also finding new devs like happened for amd64. I'm not sure if I should wait for the ATs GLEP to complete or just ask hparker of this right now, anyway that would require to have an AT operational lead for G/ALT project, right now i don't see anyone having time enough to do that... I should probably just ask for someone to step over.

Instead, being one of the few projects using the projectxml format to archive the information in a decent way, I'm missing a way to dedicate an entire page to a task. Things like non-GNU portability, sandbox and alternate prefix support are things that deserves a space on their own, but I don't feel like adding a new subproject for all of them... I think I'll ask Paul about this tomorrow, too :)

Eh tomorrow is going to be a long day, I'm afraid, a lot of documentation to write, people to talk about, bugs to smash, flames to accept (yeah I know I'll be flamed for the new Gentoo/ALT requests on gentoo-dev), and it is also Monday!

Has someone already invented a device to skip a day and pass directly to the next one? :)