I'm a trainee...

[This post is a last post via UEC network.
Tomorrow, I'll packing my real stuffs.
Day after next, I'll do moving house.
At 4/3, the first work.]

では,修行に入る. :P
[OKay, I'm a trainee. :P]

[I've take myself for ceremonial screenshot.]



勿論,違う所で. :P
[So, OLD PARR will be dead.
In the future, if I could connect WWW, accustomed myself to work,
and then, if I had good idea or so, I'll planning to re-start PARR.
Of cource, the URI is moved. :P]

ここは残るハズ,多分. :P
[Well, this blog can be continue, maybe. :P]


Some Funny GGI Folks

5時間分のログだったりする. :P
[Today's post comes from #ggi@FreeNode.net.
It's 5h log. :P]

[MooZ comes.]
3/28 01:59:44 *
MooZ (i=MooZ@FR-CHA-C4-01-02-087231004236.chello.fr) has joined #ggi
3/28 02:01:40 neiljp-work
3/28 02:01:46 MooZ
何だ何だ? :D
[Hehe, what does it start? :D]
3/28 02:02:07 neiljp-work
MooZ: am I misremembering, or where you involved in setting up the wiki?
3/28 02:02:14 neiljp-work
3/28 02:02:28 MooZ
3/28 02:02:33 MooZ
3/28 02:02:41 MooZ
i don't remember
3/28 02:02:42 neiljp-work
sounds like no ;)
3/28 02:02:55 MooZ
timlarson_ set one years ago
3/28 02:02:59 MooZ
but it got spammed :(
3/28 02:03:03 neiljp-work
3/28 02:03:18 neiljp-work
modern ones have user accounts, admin, etc
3/28 02:03:37 neiljp-work
merging with a cms, I guess
3/28 02:04:26 MooZ
i used dokuwiki and it's great and easy to install
3/28 02:04:43 neiljp-work
I use MoinMoin
3/28 02:04:51 neiljp-work
its python-tastic! ;)
3/28 02:05:01 MooZ
3/28 02:05:36 MooZ
it'ld be great to link user auth with sf one
3/28 02:05:48 MooZ
dunno if sf has some ldap or whatever for user auth
3/28 02:06:53 neiljp-work
well there was an interesting wiki idea being developed
by someone on debian-planet,
which was based on svn or similar
3/28 02:07:06 neiljp-work
so much like we do here, you could check-in pages
3/28 02:07:17 neiljp-work
on that note, of course, its not much improvement over what we have now
3/28 02:07:54 neiljp-work
I suppose I'm just keen to see more published to-do lists,
who is doing what, etc.
3/28 02:08:12 *
neiljp-work shrugs
自分の場合,増える一方だったりしますが. :P
[Yep, to list TODO is too important for me, too.
But, my TODO has only enlarging process. :P]
3/28 02:08:38 hiyuh
3/28 02:09:00 neiljp-work
hiyuh: XGGI working yet?
[In a split second, neiljp pokes me.]
3/28 02:09:38 hiyuh
XGGI seems work here. But, I have no jp106 keymap :P
3/28 02:09:51 MooZ
neiljp, well if you have python binding for cvs,
maybe you could hack to commit pages
3/28 02:10:11 neiljp-work
MooZ: sounds...dangerous!
3/28 02:10:22 MooZ
3/28 02:10:44 neiljp-work
hang on, what did you mean to connect the python to?
3/28 02:11:06 neiljp-work
hiyuh: but normal X works?
3/28 02:11:43 neiljp-work
hiyuh: have you discussed this before?
3/28 02:12:16 MooZ
neiljp-work, yeah you commit using your log/pass all via https
3/28 02:12:18 hiyuh
neiljp-work: yeah, Xorg, XGGI, XDirectFB work on same my PowerBook,
not chrooted env. :9
3/28 02:12:38 neiljp-work
hiyuh: any speed comparison? :)
3/28 02:12:51 neiljp-work
hiyuh: although that depends on the backends (targets) I guess...
3/28 02:13:26 neiljp-work
MooZ: you mean tie it to sf auth db, and then to a wiki?
3/28 02:13:38 MooZ
neiljp, yes
でも,ちゃんとテスト出来る程設定してない... :(
[So, the important status of X server is its speed, as you know.
But, I'm not OK to test it because of it's not fully prepared... :(]
3/28 02:14:19 hiyuh
neiljp-work: hmm, I've not test its speed.
because of I'm not sure to test its speed.
to use glxgears or so?
3/28 02:15:25 neiljp-work
hiyuh: glxgears goes to X glx/dri extension afaik,
which I doubt will work with xggi!
3/28 02:16:44 hiyuh
neiljp-work: heh. then, what can I use for its test? :9
jp106のキーマップファイルを作らなければならない自体に追い込まれる. :P
[Well, should I make jpp106 keymap file? :P
FYI, bughunter looks like one of the project leads.]
3/28 02:17:02 bughunter
hiyuh: Do you need some guidance on how to create a jp106 keymap ?
3/28 02:19:12 hiyuh
bughunter: maybe, I need it.
but to include how-to-create-your-keymap.txt in XGGI tarball is best, IMHO.
3/28 02:21:41 neiljp-work
I'm assuming this is a libgii issue rather than an x config issue?
3/28 02:22:13 bughunter
hiyuh: In each xggimap.* file is a line
'See the xmodmap(1) manpage for syntax explanation.'
3/28 02:22:22 bughunter
hiyuh: So first step is to read this.
3/28 02:23:17 bughunter
hiyuh: Second step is to figure out the keycode numbers to associate
with which symbol
3/28 02:23:41 bughunter
hiyuh: xev tells you this when you press a key
3/28 02:24:16 bughunter
hiyuh: Third step is to repeat second step until you're done and happy.
[Hmm, sounds like step-by-step.]
3/28 02:24:26 neiljp-work
bughunter: so this is a standard X configuration?
if so, then hiyuh probably has a keymap file for his current X server, no?
3/28 02:24:39 neiljp-work
s/configuration/configuration issue/
3/28 02:24:41 bughunter
hiyuh: You don't need to restart XGGI every time you modify your keymap file
3/28 02:25:38 bughunter
hiyuh: You can apply your modified keymap file with 'xmodmap '
3/28 02:26:12
xev probably returns different keys in xggi
compared to . Or?
3/28 02:32:52 hiyuh
neiljp-work: yeah, I have the jp106 keymap file for standard X server, IIRC.
So, in Xorg or XDirectFB, my PowerBook's keyboard works fine.
3/28 02:38:34 neiljp-work
hiyuh: but the same file doesn't work for xggi?
3/28 02:42:30 hiyuh
neiljp-work: hmm, I've not test it.
I ran XGGI by this wrapper.
3/28 02:42:36 hiyuh
maybe, somly to do it may be sane way, though.
3/28 02:42:48 hiyuh
うむむ,typo. :S
[Umm, I've typo-ed. :S]
3/28 02:52:40 neiljp-work
hiyuh: I don't currently know enough about xmodmap,
or X generally, to know how keymaps work
- one of the 'standard' keymaps is all I use for (non-xggi) X
3/28 02:52:58 *
neiljp-work admits quietly to not testing xggi
and waits for the ear-bashing from bughunter
やはりリーダーは強いな. :)
[Um, any leads must be tough. :)]
3/28 02:59:52 hiyuh
neiljp-work: hmm, maybe, I use it for jp106 keymap symbols with standerd X.
it looks like not for xmodmap.
3/28 03:00:58 hiyuh
neiljp-work: it's provided by xkbdata as one of Xorg stuffs, IIRC.
3/28 03:01:11 neiljp-work
ah, hang on
3/28 03:04:26 neiljp-work
hiyuh: xorg.conf or whatever normally specifies the keymap (right?),
but xggi doesn't have that
- somehow xggi must find which keymap to load,
so can you use some option to specify that keymap file you have already?
3/28 03:04:38 neiljp-work
hiyuh: I don't have time to investigate in detail,
just chat in between other things :)
3/28 03:09:41 hiyuh
neiljp-work: yeah, my xorg.conf specfies jp106 keymap.
please see line 87.
it's the option for jp106 symbols, IIRC.
3/28 03:11:47 neiljp-work
hiyuh: yes, so how does xggi specify the keymap? :)
I don't have the xggi docs to hand, so that's over to you :)
3/28 03:11:54 neiljp-work
(that or else bughunter said it above ^^)
3/28 03:14:32 hiyuh
yeah, btw, startxggi.sh in my wrapper which I pasted
is just renamed startx.sh which is bandled in XGGI tarball.
and, I've skimed it.
IIRC, it executes just xmodmap with xggimap.base or so.
3/28 03:16:42 hiyuh
so, anyway, I need to create my jp106 keymap file for XGGI
as a xmodmap file like bughunter said. :)
3/28 03:18:16 hiyuh
Otherwise, DEL/BS key does not work. :P
そう,一番困るのがDEL/BSキーが効かない事なのです. :(
[So, my mainly PITA is DEL/BS key doesn't work. :(]
3/28 03:22:48 neiljp-work
too much to hope that a keymap file is an xmodmap file? ;)
3/28 03:23:28 neiljp-work
hiyuh: maybe xmodmap -pke in a working xserver should generate your xmodmap file?
3/28 03:23:37 neiljp-work
that is, if those files are different
3/28 03:25:15 hiyuh
ZOMG, it seems genarate my xmodmap file! :9
なんてこった,いきなり自動生成かよ?! o.O
[WTF, it looks like a automatically generating magic?! o.O]
3/28 03:25:34 neiljp-work
ah man pages, aren't they great? :)
3/28 03:26:10 hiyuh
exactly. ;)
3/28 03:26:23 hiyuh
3/28 03:26:45 neiljp-work
of course, I'm intrigued by whether xggi could just use standard keymap files
like xorg/xf
やはり,man pagesは重要ですねー. :9
[Exactly, man pages are TOO informative. :9]
3/28 03:27:16 neiljp-work
I'll build xggi sometime,
as opposed to now when I'm supposed to be working on a presentation for tomorrow...
3/28 03:30:13 hiyuh
3/28 03:31:30 hiyuh
So, XGGI is fully prepare for me.
3/28 03:32:07 neiljp-work
it works ok?
3/28 03:32:28 neiljp-work
now you can benchmark ;)
3/28 03:32:43 neiljp-work
that or test display-multi and other fun :)
3/28 03:33:06 neiljp-work
or whatever else you find fun for that matter...
3/28 03:34:56 hiyuh
neiljp-work: could be, then I'll switch to XGGI. bbiab ;)
[Then, I've left the channel.
And tweak my XGGI wrapper or so...]
3/28 04:31:45 hiyuh
hmm, doesn't work :(
3/28 04:40:51 *
hiyuh is testing his key{board,code} with xev...
3/28 04:48:49 hiyuh
btw, XGGI can be built with mesa?
動かん... :(
仕方無いので,チョコチョコやる事にする. :S
[But, unfortunately, it doesn't work. :(
And then, I've asked about XGGI+Mesa status.]
3/28 05:10:37 neiljp
hiyuh: what does/doesn't work?
3/28 05:11:48 bughunter
hiyuh: It may, it may not.
It doesn't make a difference since XGGI has no GLX extension yet
3/28 05:26:31 hiyuh
neiljp: I've generated it by xmodmap -pke like you said.
and then, I've used it with/without xggimap.base by xmodmap
(or, it requres some tweaks?).
but some keys get more messy. (e.g. return key does not work or so.)
3/28 05:30:02 hiyuh
bughunter: ok, thanks.
if so, to build with mesa makes only compiling time longer, ATM.
it's nonsense, IMHO.
3/28 05:32:54 neiljp
hiyuh: stupid question, but does xggi work in x? :)
3/28 05:33:01 neiljp
hiyuh: ie. all the keys, etc
3/28 05:33:14 neiljp
hiyuh: that is, assuming you're running xggi only on fbdev atm
3/28 05:33:23 neiljp
(from your script)
なぬ? o.O
Xnestか? 8-)
[Eh? o.O
Well, it looks like Xnest? 8-)]
3/28 05:41:37 hiyuh
neiljp: well, you mean XGGI can run in other X?
3/28 05:42:06 neiljp
hiyuh: I assume so, if you set the GGI targets correctly
3/28 05:42:19 bughunter
hiyuh: yes, it can. This way you can debug XGGI with ddd, for example.
3/28 05:43:06 hiyuh
heh, sounds like Xnest.
3/28 05:43:43 bughunter
yeah, but can you in turn run Xnest in Xnest ?
3/28 05:43:46 neiljp
of course if the bug is in linux-kbd
or whatever input you have on the console with fbdev,
then it won't help debugging that
3/28 05:43:57 bughunter
you can run XGGI in XGGI in XGGI in Xorg, for example
3/28 05:44:21 bughunter
so who cares about Xnest? ;-)
3/28 05:44:27 hiyuh
そこまで言うか,流石. ;)
で,新しい設定をチェックして頂く. :9
[Hehe, looks like so aggresive. ;)
Then, I'd like to check my new configurations. :9]
3/28 05:52:30 hiyuh
So, like GGI_DEFMOD="640x480[C24]" GGI_DISPLAY="display-x:-noinput"?
3/28 05:52:40 bughunter
3/28 05:52:53 hiyuh
bah, sorry. :(
3/28 05:55:14 bughunter
and what do you use as input?
3/28 05:58:18 hiyuh
I've set it as GGI_INPUT="(input-linux-mouse:imps2,/dev/input/mice):(input-linux-kbd)"
3/28 05:58:27 bughunter
3/28 05:58:37 bughunter
3/28 05:59:02 bughunter
I think, this GGI_INPUT is ok when you want to run XGGI on fbdev, for example
3/28 05:59:33 bughunter
but to run XGGI on X, it is better to use the native display-x input
3/28 06:00:02 bughunter
that means, unset GGI_INPUT and set GGI_DISPLAY="display-x"
3/28 06:00:30 hiyuh
yes, sir. :9
3/28 06:09:24 hiyuh
bughunter: so, like this? http://rafb.net/paste/results/FzPqqX29.html
3/28 06:10:58 neiljp
talking about nesting xggi made me think of a problem I had recently
- my x server crashed :(
3/28 06:11:50 hiyuh
oh, well... it's horrible. :P
3/28 06:12:08 bughunter
hiyuh: what is startxggi.sh ?
3/28 06:12:14 neiljp
could xggi be used as a proxy?
so clients connect to xggi, and it then connects to a 'real' x server?
3/28 06:12:57 bughunter
3/28 06:13:55 neiljp
ideally xclient<->xggi<->xorg, but x connections don't like breakages, right?
ie. if xorg dies, xggi dies too
3/28 06:14:21 neiljp
so you need the ability to disconnect the chain between xggi and xorg
3/28 06:14:24 bughunter
hiyuh: you need to tell XGGI somewhere which display to use
- i.e. if x.org runs on :0, then XGGI must run on :1, etc.
3/28 06:15:11 bughunter
hiyuh: and then your x apps must connect to display :1 to run within XGGI.
3/28 06:16:10 neiljp
xvnc is one way to disconnect the chain,
ie. I guess: xclient<->xggi<->xvnc<===> xvncclient<->xorg
3/28 06:16:14 bughunter
neiljp: You want Xserver-balancing ? :)
3/28 06:16:52 bughunter
xvncclient<->xggi<->fbdev :)
3/28 06:17:00 neiljp
when I was involved in fresco, and chatting to other projects too,
there was the idea of being able to move your display to
for example a mobile device
3/28 06:17:11 neiljp
bughunter: sure, but your apps keep running
3/28 06:18:03 neiljp
of course, xvnc is effectively a server in there too,
which isn't really needed if you don't want to use vnc protocol
3/28 06:18:26 neiljp
if you only care about local, it would be useful to have xclient<->xggi<->memory<->xorg
3/28 06:19:23 neiljp
then we effectively delegate xorg to what it 'should' be...
a graphics driver display framework ;)
3/28 06:19:51 neiljp
ie. xggi handles xlib and xorg handles rendering
3/28 06:21:02 neiljp
the problem of course is handling the rendering requests
so that xorg rendering is accelerated via the xggi/memory requests
3/28 06:21:43 *
neiljp wonders if bughunter is whipping up some clever app in the background ;)
3/28 06:22:22 neiljp
anyhow I've emptied my brain of that, so time for me to get back to some ironing :|
3/28 06:22:40 hiyuh
bughunter: startxggi.sh is equal to startx.sh (I've renamed it)
which was one of xggi-2.0.0.src.tar.gz stuffs.
3/28 06:27:17 bughunter
hiyuh: ok. Then pass XGGI's display number as argument to startxggi.sh
3/28 06:27:30 bughunter
hiyuh: /usr/bin/startxggi.sh :1 (for example)
3/28 06:27:46 bughunter
hiyuh: It is recommended to use the next free display number
3/28 06:28:07 bughunter
hiyuh: (see my comments above)
3/28 06:28:19 hiyuh
okay, I'll try that.
と言う事で,またムニャる. :S
[Then, Munya-Mnya. :S]
3/28 06:33:17 neiljp
bughunter: I suppose that what I am after is disconnect/reconnect operation
for ggi targets
3/28 06:33:30 neiljp
or at least display-x
3/28 06:36:25 bughunter
or more generic: let GGI apps attach/detach
to display-targets at runtime very similar to screen
3/28 06:36:35 neiljp
a hypothetical situation: GGI_DISPLAY=x ggi_app;
[ggi_app connects to DISPLAY using display-x]
[send a message to ggi_app, the ggi part of it, GGI_DISPLAY=x:screen=:1]
[ggi unloads x target, reloads x target with DISPLAY=:1]
3/28 06:36:37 bughunter
neiljp: Is that right?
3/28 06:37:04 neiljp
3/28 06:37:05 neiljp
3/28 06:37:37 neiljp
of course, then you need to handle input too...
3/28 06:37:51 neiljp
but that's my crazy idea for the day/week/month :)
3/28 06:37:51 bughunter
3/28 06:37:57 neiljp
3/28 06:38:31 bughunter
.oO(We should add a 'Idea of the day' to our GGI homepage)
3/28 06:38:53 bughunter
or perhaps 'Quote of the day'
3/28 06:39:59 bughunter
something like 'Blackholes are where god divided by zero'
3/28 06:40:04 bughunter
3/28 06:40:44 neiljp
atm you could do: xclient<->xggi<->memory, hacky_app(memory)<->xorg
3/28 06:40:55 neiljp
which would be like a shortcut vnc
3/28 06:41:07 bughunter
another possible 'Quote of the day':
3/28 06:41:08 bughunter
"By filing this bug you have questioned my family honor. Prepare to die!"
3/28 06:41:08 bughunter
-- Klingon programming - rule 8
3/28 06:41:13 bughunter
3/28 06:41:15 neiljp
or ggiteleserver would work
3/28 06:41:22 neiljp
3/28 06:41:30 neiljp
methinks bughunter isn't taking my idea seriously ;)
3/28 06:42:07 bughunter
well for now we have to bring -current back into working state
3/28 06:42:21 bughunter
then we can think on what's next
3/28 06:44:29 bughunter
"A bus station is where a bus stops.
3/28 06:44:29 bughunter
A train station is where a train stops.
3/28 06:44:29 bughunter
Now, on my desk, I have a work station..."
3/28 06:45:32 hiyuh
マジメに話しているのかと思ったら,何をやっとるんだ,アンタ等. :D
[IMAO, ya must be funny folks. :D]

あ,因みにまだ動いてませんよ,XGGI on Xは. :P
[Well, XGGI on X does not run by my configurations, ATM. :P
All I can say about it is:
It must be set acceptable ${DISPLAY} variable.
Otherwise, it doesn't work.]


gensplash + qingy

[Well, it's a followup.]

bootsplashとgensplashは別物.参考:spock's devsite gensplash::FAQ 1
[bootsplash isn't gensplash. see also: spock's devsite gensplash::FAQ 1
--gensplash=${THEME} as a genkernel's option requires directroy's name which is in under the /etc/splash.
--gensplash-res=${WIDTH}x${HEIGHT} as a genkernel's option requires to set sane resolution of your display, and then make sure that you have /etc/splash/${THEME}/${WIDTH}x${HEIGHT}.cfg.]

If you want to use genkernel, to use emergence as a theme of gensplash, your display resolution is 1024x768, then please do like that.]
 $ genkernel --menuconfig --gensplash=emergence --gensplash-res=1024x768 all
[So, --bootsplash as a option of genkernel is NOT mandatory with it.
When configuration the kernel, make sure to set CONFIG_FB_SPLASH=y.
If you already had like that setting, all you have to do is,]
 $ genkernel --oldconfig --gensplash=emergence --gensplash-res=1024x768 initrd
[for re-generate your initramfs.]

[Of cource, your kernel sources were patched with the fbsplash patch before you built the kenrel/initramfs.
sys-kernel/gentoo-sources seems be patched when it released.
If you use other kernel sources, you have to confirm whether it's patched.
Then, after to generate initramfs succesfully, you'll see some warning for using the gensplash with some boot-loader.
It's as known as kernel parameters for boot your kernel, so you have to set it.]

[Next is sys-apps/qingy.
It requires dev-libs/DirectFB, so some dangerousness are.
Like I said, dev-libs/DirectFB sometimes breaks its backward-comptibilities,
so some apps were linked with it will lock-up/break.
And then, it can use fusion as a kernel module.
So, If you forgot to re-compile it when upgrade the kernel,
your system will lock-up/break.]

new ebuild: fusion for DirectFBを見てください.
[Unfortunatlly, the fusion module is not contained in official portage tree at the moment.
So, it means that emerging dev-libs/DirectFB with USE="fusion" will have compilation failure at the moment.
If you want to emerge it with USE="fusion", please see new ebuild: fusion for DirectFB.
However, if you want to use dev-libs/DirectFB ONLY for sys-apps/qingy, it's NOT mandatory to emerge it with USE="fusion".
At the moment, the fusion module is ONLY required by x11-base/{xdirectfb,XDirectFB}.
And then, fusion as a USE flag may be masked by some profiles' use.mask for many packages' stabilities.
Of cource, the fusion module works fine on many architectures, IIRC.]

/etc/make.confのUSEフラグに-directfb -fusionを入れておいて,
dev-libs/DirectFBにはUSE="fbcon gif png jpeg truetype zlib sysfs"が有れば,
辺りを適当に弄って下さい. :P
[I recommend to set USE="-directfb -fusion" in /etc/make.conf.
And, if you want to use the app with dev-libs/DirectFB, set directfb as a USE flag in /etc/portage/package.use.
At least, all you have to do for using sys-apps/qingy with dev-libs/DirectFB is to emerge dev-libs/DirectFB with USE="fbcon gif png jpeg truetype zlib sysfs", IIRC.
After emerge it, feel free to modify /etc/inittab, /etc/qingy, /use/share/qingy or so, AT YOUR OWN RISK. :P]


GentooJP.Nomikai++; // OSC2006 Spring

さっきまで,死んでた. :P
[Well, I've slept just now. :P]

[Friday, I atedded OSC.
The turnout of OSC at the friday seems smaller than Saturday's.
We (I and Kalin) have talk with some gentoo users who atended OSC
"To talk about gentoo on IRC w/ STEP3 is maybe OK",
"~ARCH on gentoo is so fun",
"Well, where is Gentoo on OpenBlocks?",
"emerge on poor hardware is too dull",
"I recommend to use glsa-check", or so on.
After we closed gentoo booth at about 17:00,
I talked with Kalin at DOUTOR who is Gentoo evangelist.
He said like that.
"My servers are everything's ready, for gentoo."
"The first ebuild of bluetooth-alsa is, well, subtle.", or so on.]

遅刻してしまってごめんなさい. :(
Lightning Talkの為,Kの人は不在でサインはもらえませんでした,残念.:S
「Kernel Trafficはどーなっとるんだ?」とか,色々.
[Saturday, I attended OSC, too.
I had set my alarm at 7:00, but I've overslept. :(
The turnout of OSC on saturday seems larger than Friday's.
I've get the book at OSASK booth, the guy of "K" isn't at the booth.
Because of he had to prepare for lightning talk. :(
Then, Gentoolists went to Nomikai.
We talked about like that.
"groff's -Tnippon as a option is not good idea.",
"In fact, the documents are only written in English is good?",
"well, SpanKY++.",
"Kernel Traffic breaks?", or so on.]

敢えて,誰が誰とは言わない. :P
[Here are some pictures of OSC thanks to Kalin.
Well, I can't say who is who. :P]


mixed stuffs related gentoo #2

整理するのが面倒になってきた. :P
[Hmm, to separate some stuffs is so dull for me. :P
These are my DONE/TODO list related gentoo...]

 To make/contribute ebuilds for XGGI 2.0, GGI 2.2
  See bug #126400 in Gentoo bugzilla.
 To use wmii current via hg tip via ebuild
  See my portage overlay.
 For GentooJP exhibit at OSC2006 spring
  Cleaning my 17"LCD
  To prepare some UTP cables, a tap for power supply, NETGEAR HUB

...怪しい. :P
[Umm, It's non-related gentoo thing.
I'm doing some formalities for my house moving.
My new room is in nearby O-miya.
The room is 3rd floor of a bldg.
It has a wooden floor of 6-mat and a tatami floor of 6-mat.
It has also a bath room, a rest room, a kitchen, with air conditioner.
I didn't understant why it has like a closet which is for a laundry machine.
The cost of per month is about 50,000 yen.
... Well, sounds like fishy. :P]

野宿? :P
[I'll do my house moving at 4/1.
But, the guys said to me like that:
"Please, prepare to take over the room for new student until 3/31."
WTF, rough sleeping? :P]


My planning of OSC2006

Archive of GentooJP misc
と言う事で,誰か来て下さい. :)
[So, come on OSC2006. :)]


mixed stuffs related gentoo

[These entries are mixed stuffs.]

先ず,gcc for PICを見付けたので,
crossdev guruなSpanKYにちょっかいを出した.
[first, I had reported gcc for PIC to SpanKY aka the crossdev guru.]
 2/28 21:06:55 hiyuh
SpanKY: around?
2/28 21:20:46 hiyuh
SpanKY: It may be a mere/useless info for you, though.
seems binutils/gcc for Microchip PIC port.
I know dev-embedded/sdcc already supports it.
but, crossdev supports Atmel AVR, PIC and AVR are simlar MPU.
So, corssdev can/will be supported it?
2/28 21:21:55 *
hiyuh couldn't understand toolchain-binutils.eclass
2/28 21:42:09 brix
PIC support in crossdev would be nice
2/28 23:44:33 SpanKY
hiyuh: is the code integrated into official binutils/gcc sources ?
2/28 23:51:52 hiyuh
SpanKY: maybe, it's unofficial one or a fork, I guess.
that's why its sources are only available from Microchip's web page.
ま,自分も既に使ってないから,放置. :P
[Hum, it didn't sound kinda like.
It's unofficial port and PIC is not popular as it is in Japan.
Anyway, I don't touch PIC now, so I'll ignore it, though. :P ]

これでArmadillo-9はembedded centricなブツになる,ハズ. :P
I would to build arm-unknown-linux-uclibc toolchain from
So, my armadilli-9 will be embedded centric by this toolchain. :P
 3/01 17:51:01 hiyuh
SpanKY or psm: around?
3/01 17:52:33 hiyuh
I had uclibc's build process failure
when use crossdev's arm-unknown-linux-uclibc,
it's odd.
3/01 17:53:28 hiyuh
see, this.
it failed non-related /etc stuff are readed.
3/01 17:54:06 hiyuh
but, on pure console. build successfully, like that.
3/01 17:54:53 hiyuh
should I file it as a bug?
3/01 17:55:41 psm
arm can't build it's locales, you have to use USE=pregen
3/01 17:56:49 psm
the locales support in ebuild is incorrect,
so you are on your own if you enabled it
3/01 17:57:43 hiyuh
um? then, why uclibc had build successfully
w/ USE="-pregen" on pure console?
I didn't change any setting, though.
3/01 17:59:30 psm
what is "pure console"?
3/01 17:59:44 hiyuh
it means no X.
3/01 18:00:21 psm
arm can't build it's locales if native/host system is uclibc
3/01 18:01:58 psm
what is /etc/wmii-2.5/extern?
3/01 18:02:21 hiyuh
it a wm config file.
3/01 18:02:56 hiyuh
3/01 18:04:58 hiyuh
so, it's completely non-related uclibc file.
I can't understand why it was readed by uclibc build process...
3/01 18:13:34 hiyuh
psm: the correction means
"if it set CTARGET=arm-*-uclibc and USE="-pregen"
on CBUILD=*-uclibc box,
crossdev will fails to build any locales.
that's why the part of its locales build in uclibc's ebuild
in portage is broken.", right?
3/01 18:15:01 psm
the ebuild is broken generally,
if you dont use pregen, independently of arch
3/01 18:15:31 hiyuh
heh, didn't know that.
3/01 18:15:45 psm
currently only mips/ppc/x86 can rebuild
on an uClibc native host the locales
3/01 18:18:05 psm
it seems that awk gets wrong for print $0
(see end of extra/locale/Makefile), getting that /etc/wmii...
3/01 18:19:00 psm
what awk do you use? gawk?
could you try mawk (change the awk link from gawk to mawk)
3/01 18:20:40 hiyuh
on X, "which awk" said /usr/bin/awk,
than "equery belongs /usr/bin/awk" said sys-apps/gawk-3.1.5
3/01 18:20:54 hiyuh
3/01 18:21:05 psm
do not reemerge, go only to extra/locale,
run make clean, make
3/01 18:21:43 psm
you have one in /bin/ as well probably
3/01 18:22:18 hiyuh
yeah, /bin/awk is.
3/01 18:22:56 psm
even pregen won't help w/ the /etc/wmii... part running on X,
the command is the same for pregen and own-locale-build
3/01 18:24:34 psm
try mawk, let's see if that gets it right,
if not the script has to corrected, I can't though,
I do not know awk enough
3/01 18:25:34 psm
az: ping
3/01 18:27:21 hiyuh
well, though, mine box is ppc, but its libc is glibc.
so, to build uclibc chroot enviromnent
for arm development is better than
to install arm's cross-compiler directly to /?
it was installed / directlty for just test, atm.
3/01 18:28:49 psm
I wouldn't do cross-compile ppc-glibc to arm-uclibc,
better if need it, ppc-uclibc chroot, there arm-cross
3/01 18:30:26 hiyuh
and then, it is a chance to try hardend? :P
3/01 18:30:46 psm
you are crazy ;)
或る意味,最高の褒め言葉だね,それは. ;)
[Yeah, it sound kinda like good word for me. ;)]
 3/01 18:31:47 az
psm: sup?
3/01 18:31:48 psm
I would build hardened only native,
cross-hardened compiling is not tested
3/01 18:33:07 psm
az: could you take a look at uClibc-0.9.28/extra/locale/Makefile?
the last line where awk is used gets instead of $0 something from X
(hiyuh can tell where that comes from, see his earlier paste)
3/01 18:33:47 az
mind pasting that bit ?
3/01 18:34:23 psm
ok, a mom
3/01 18:36:34 psm
az: http://rafb.net/paste/results/ewnB7L35.html
3/01 18:38:36 hiyuh
az, psm: well, complete uclibc's build log is.
should I paset is?
3/01 18:39:03 hiyuh
3/01 18:39:24 psm
hiyuh: where is /etc/wmii-2.5/extern occuring in your environment?
3/01 18:40:17 az
uhm, so it prints whatever until it gets to /WANT_/ ?
3/01 18:41:03 hiyuh
psm, maybe, its setting comes from ${HOME}/.wmii-2.5/wmiirc.
3/01 18:43:59 hiyuh
it's complete uclibc's build log which it failed on X.
3/01 18:44:15 az
anyhow weird way to do it
3/01 18:44:50 az
why not just:
cat uClibc_locale_data.h |
awk '{ if ( /WANT_/ ) exit ; else print }' >
3/01 18:45:52 az
which could be optimised by not using UUOC
3/01 18:46:07 az
awk '{ if ( /WANT_/ ) exit ; else print }' uClibc_locale_data.h >
3/01 18:46:54 psm
az: thanks, I will check if the resulting file is the same
3/01 18:48:42 az
problem might be that 'i' is intialized as a int,
and then set to a string later on
3/01 18:48:53 az
gawk should handle it fine, but not sure about others
3/01 18:51:01 az
hmm, maybe not .. its set to the regex evaluation
3/01 18:51:01 hiyuh
hmm, I'll re-setup my development environment
in chroot uclibc userland.
and report it if it need.
3/01 18:51:44 psm
gawk was the one failing/replacing $0 from user env
3/01 18:55:16 az
it really should not care about environment variables,
except if set via -v
3/01 18:55:29 az
anyhow, what i pasted wont work
3/01 18:55:42 az
that is to strip WANT_ ifdefs from the code
3/01 18:57:22 hiyuh
ok, I'll try to use mawk and uclibc userland, though.
like psm said,
ppc-uclibc to arm-uclibc is better than ppc-glibc to arm-uclibc,
I think, too.
3/01 18:57:52 hiyuh
az, psm: anyway, thanks. :)
3/01 18:58:19 az
can you run that manually an check on what line it barfs ?
3/01 18:59:08 psm
az: your pastes?
3/01 19:00:19 psm
the generated file is the content between ifdef/endif of
the earlier that has many #ifdef WANT_something #endif ...
3/01 19:01:59 az
generated is everything outside #ifdef WANT_something #endif
if im not mistaken
3/01 19:02:16 psm
3/01 19:02:31 az
could try to do it a bit differently by setting i explicit,
and not via using regex result:
awk 'BEGIN{ i=1 } { if ( /WANT_/ ) i = 0 ; else if (i) print $0;
if ( /endif.*WANT_/ ) i = 1 }'
3/01 19:03:39 hiyuh
"ebuild cross-arm-unknown-linux-uclibc/uclibc compile"
failed now on X.
what I take it? :P
3/01 19:04:49 az
awk: (FILENAME=- FNR=1) fatal: division by zero attempted
3/01 19:05:00 az
^^ bug in awk or maybe libc ?
3/01 19:05:08 hiyuh
今考えると笑えないな,おい. :(
[Oh, well. it's nightmare rather than not funny. :(]
 3/01 19:06:23 psm
hiyuh: you should crossdev there as well,
do you have binutils/gcc-stage1/headers already?
3/01 19:09:03 hiyuh
psm: oh, well.
so, you mean that I should unistall it, and then re-run crossdev?
3/01 19:10:17 psm
hiyuh: I do not know what you did until now:
stage3-uclibc-ppc/chroot/crossdev -s2(at least) arm-*-linux-uclibc?
3/01 19:15:07 hiyuh
psm: ah, I had run crossdev on console
for build arm-unknown-linux-uclibc.
because of ppc-glibc to arm-uclibc corssdev failed like I pasted.
so, anyway it's installed into / directly atm.
and then,
I did "ebuild cross-arm-unknown-linux-uclibc/uclibc compile"
for testing. it's not enough to test it?
3/01 19:19:56 psm
you can reuse crossdev ,
it will not rebuild binutils/gcc if they already there
3/01 19:20:16 psm
you could try az's change for extra/locale/Makefile
if it works for you
3/01 19:22:55 hiyuh
but, uclibc is alreadly installed
that's why crossdev had no error on console.
so, just re-run crossdev do nothing atm.
now, only to uninstall uclibc is required at least, wrong?
3/01 19:23:35 psm
remove the cross uclibc then and rerun crossdev
3/01 19:23:51 hiyuh
ok, I'll try that. :9
3/01 19:30:56 hiyuh
ok, I did so. then edit extra/locale/Makefile?
3/01 19:32:20 hiyuh
hmm, links-target as a target section?
3/01 19:34:26 hiyuh
well, pregen as a target has same awk script, too.
で,いきなりネットワークの調子が悪くなる. :/
[My stupid connection was dead atm. :/]
 3/01 19:50:04 hiyuh
sorry, my stupid connection. :(
3/01 19:53:00 hiyuh
psm: like this? http://rafb.net/paste/results/dEkDBK90.html
英検3級しか持ってないもんな. :P
[While it has dead connection,
2006.0 release announcement are translated by my stupid english skill.
It'll be completely re-written by GentooJP doc team, though.
Because of, I have had no TOEIC/TOEFL, I have had only SETP-3. :P]
 3/01 22:44:43 hiyuh
psm: sorry, about my network connection was dead.
btw, you have seen my diff for extra/locales/Makefile's diff at rafb?
3/01 22:46:57 psm
3/01 22:48:28 hiyuh
ok, please see this. http://rafb.net/paste/results/oNdt8S35.html
3/01 22:49:42 psm
does it work for you under X?
I have tested the output and it is ok (will check w/ mawk as well)
3/01 22:50:19 hiyuh
I haven't test it, though. :P
3/01 22:52:09 hiyuh
psm: then, "make" at the top of uclibc dir, right?
or epatch in ebuild is better?
3/01 22:59:13 psm
if using 0.9.28, then rm -f include/bits/uClibc_locale_data.h;
cd extra/locale; make
3/01 23:03:50 hiyuh
psm: thanks, I'll try that.
3/01 23:05:45 hiyuh
psm: ouch, I get same error. :(
3/01 23:07:08 psm
hiyuh: set | grep wmii ?
3/01 23:09:56 hiyuh
psm: didn't know that.
its result has three entries, wtf.
3/01 23:11:33 psm
hiyuh: why is 0 set?
that is the one catched by $0, but why only within X?
3/01 23:12:29 psm
arm is used as well,
maybe some wants to make a benefit by supporting those
3/01 23:14:25 hiyuh
psm: yeah, wmii is a bit of strage from normal wm, IMHO.
it is featuring Plan9. maybe, it cuz my uclibc build failure problem.
3/01 23:15:37 psm
Does this 0 mean display 0?
3/01 23:16:50 psm
or :x.0 maybe?
3/01 23:18:32 hiyuh
maybe, no.
/etc/wmii-2.5/extern is a 9base's rc script
which comes from utils Plan9 ports.
it controls keyboard short-cut feature or so, IMHO.
3/01 23:20:01 psm
does the wm itself set 0, or some script?
3/01 23:22:08 hiyuh
psm: maybe, configurable.
3/01 23:22:49 hiyuh
so, X might have no problem,
but my wm is wmii by default atm.
if I had build chroot uclibc userland
and when I use it in wmii by this setting,
same problem will happen, IMHO.
3/01 23:28:04 hiyuh
psm: btw, =wmii-2.5.* depends on 9base's rc script backend.
but, >=wmii-3 will not depend on it, it use *NIX sh backend, I heard.
3/01 23:29:09 hiyuh
psm: so, uclibc's Makefile does not require any change.
3/01 23:30:41 hiyuh
psm: anyway, thanks. I can understand this problem, now. :)
3/01 23:45:33 psm
az: what do you think of it,
the env has 0 set to something and Makefile/awk gets it
3/01 23:46:14 az
the $0 awk uses should have nothing to do with env variables,
except if its used in the code in the Makefile and not escaped properly
3/01 23:46:40 az
ie, in the Makefile it will be $$0 and not $0
3/01 23:47:38 *
hiyuh sees some mails
about "wmii will use sh backend instead of 9base rc script" in its ml.
it's kinda like flame war.
x11-wm/wmii-3*が必要っぽい. :(
[Now, it's clear, it cuz sys-apps/9base rc script for wmii-2.5.
Umm, I need x11-wm/wmii-3, though. :(]