[00:08:00] *** Proctop has quit IRC
[00:12:18] *** LupusMichaelis has joined #haiku
[00:30:15] *** nPHYN1T3 has joined #haiku
[00:30:15] *** ConneX has quit IRC
[00:32:27] *** Begasus has joined #haiku
[00:36:05] *** LupusMichaelis has quit IRC
[00:47:15] *** ablyss has quit IRC
[00:48:53] *** Aimy|afk has quit IRC
[00:50:30] *** ablyss has joined #haiku
[01:05:40] *** RageMax has quit IRC
[01:12:44] *** MikeW has quit IRC
[01:17:43] *** RageMax has joined #haiku
[01:22:07] *** ablyss has quit IRC
[01:28:22] <[Beta]> why does some code #ifndef __BEOS__ around size_t, I presume it doesnt have it ?
[01:29:10] <slaad> Because size_t is a BeOS type.
[01:29:23] <slaad> I think...
[01:29:30] <@Dr_Evil> no
[01:29:34] <@Dr_Evil> BeOS has it
[01:30:11] <[Beta]> so..
[01:30:14] <[Beta]> +#ifdef __BEOS__
[01:30:14] <[Beta]> +			int fromlen;
[01:30:14] <[Beta]> +#else
[01:30:14] <[Beta]> size_t fromlen;
[01:30:14] <[Beta]> +#endif
[01:30:17] <[Beta]> is silly ?
[01:30:46] <slaad> Yeah, it is. size_t actually seems to be a POSIX type.
[01:31:35] *** TLF has quit IRC
[01:31:36] <@Dr_Evil> yes, very silly
[01:32:45] <[Beta]> ok
[01:33:07] <[Beta]> wish I knew why BeMonni did that change.
[01:33:43] <@Dr_Evil> which change?
[01:33:55] <@Dr_Evil> what got changed?
[01:34:16] <[Beta]> He's working on a diff to reduce compiler warnings, etc
[01:34:59] <[Beta]> <BeMonni>       http://www.raasu.org/tools/haiku/haiku-Jan-01-2005.diff <--- still needs a little cleaning up, but does tell pretty well where is issues, now with right year :)
[01:35:25] <@Dr_Evil> he should post small patches to the mailinglist, obviously he is doing stuff wrong and shouldn't waste his time
[01:35:46] <[Beta]> well - if the changes are ok, it wont be a waste of time..
[01:35:50] <@Dr_Evil> lets see:
[01:35:59] <@Dr_Evil> posix/string.h rejected, wrong way
[01:36:20] <@Dr_Evil> et6x00/Acceleration.c using static would be better
[01:36:38] <@Dr_Evil> accelerants/radeon/CP.c, using static would be better
[01:36:54] <@Dr_Evil> same for the next few files
[01:37:19] <@Dr_Evil> auvia/debug.c accepted
[01:38:10] <@Dr_Evil> auvia/multi.c,v important bugfix, should be done immediately and reported to fyysik
[01:39:21] <@Dr_Evil> usb_audio/USB_audio_utils.c  looks valid, also instead of using // the linus shoud better be removed
[01:40:01] <@Dr_Evil> radeon/CP_setup.c,v  yes, valid bugfix
[01:40:02] <[Beta]> one presumes for those (now empty) functions, they would be left as a ToDo hint.
[01:40:24] <fyysik> 
[01:40:28] <fyysik> hmm
[01:40:34] <@Dr_Evil> via-rhine/hook.c,v  thats very ugly, should be made static
[01:41:13] <@Dr_Evil> well, I'll stop now
[01:41:38] <fyysik> Dr_Evil - where is patch for multi.c?
[01:41:42] <[Beta]> thought you never would :)
[01:41:59] <@Dr_Evil> however, work could be avoided if BeMonni would stop doing changes and finally sends patches, so they can be rejected ;-)
[01:42:10] <fyysik> looking
[01:42:20] <[Beta]> line 440
[01:42:30] <@Dr_Evil> default:
[01:42:30] <@Dr_Evil> -                               strncpy(data->friendly_name, FRIENDLY_NAME);
[01:42:31] <@Dr_Evil> +                               strncpy(data->friendly_name, FRIENDLY_NAME, 32)
[01:42:46] <fyysik> hmm, will try
[01:42:54] <@Dr_Evil> I'm not sure if this can lead to a crash
[01:43:43] <@Dr_Evil> all those  strncpy calles are wrong anyway, as they don't 0 terminate the strings
[01:44:34] <@Dr_Evil> thats why I suggested a code review some days ago
[01:44:47] <@Dr_Evil> strncpy and freinds are often used in a wrong or unsave way
[01:44:59] <@Dr_Evil> and should be replaced by strlcpy and friends
[01:45:30] <slaad> It should be fine if you've memset() to 0 / calloc'd first, shouldn't it?
[01:45:38] <fyysik> sad, not the case here
[01:46:01] <fyysik> my version of multy contains 32 constant in strncopy parameter list
[01:46:10] <@Dr_Evil> slaad no
[01:46:29] <@Dr_Evil> slaad then you had to use 31 as paramter to strncpy
[01:46:50] <@Dr_Evil> else there is no 0 byte at the end, if the source string is longer than 32 bytes
[01:46:59] <@Dr_Evil> or equal to 32 bytes
[01:47:01] <slaad> But does strlcpy do that?
[01:47:07] <slaad> I'm presuming it does.
[01:47:10] <@Dr_Evil> yes
[01:47:27] * Dr_Evil should do the code review, but he is busy writing new code ;)
[01:48:41] <[Beta]> why we dont just make a function that caps the end of the var..
[01:48:50] <fyysik> but i will look for similar strncopy improper usage in other places. who knows
[01:48:53] <[Beta]> sorry, macro.
[01:49:17] <Dr_Evil> [Beta] strlcpy does so
[01:49:24] <@Dr_Evil> but it's not in R5
[01:49:59] <[Beta]> d'oh.
[01:50:11] <[Beta]> does haiku have it ?
[01:51:53] *** Striver has joined #haiku
[01:52:51] <fyysik> no strlcpy in beos headers
[01:55:48] <@Dr_Evil> haiku will have it
[01:55:56] <[Beta]> will meaning not yet ?
[01:56:04] <@Dr_Evil> haiku has it
[01:56:08] <[Beta]> ok.
[01:56:25] <@Dr_Evil> but if you compile a driver in haiku that uses strlcpy, it wont run on R5 obviously
[01:56:45] <[Beta]> so then we code to strlcpy, and have a #ifndef strlcpy ... replacement line?
[01:57:00] <[Beta]> I cant see a problem; but i'm a c/be newbie :)
[01:58:05] <@Dr_Evil> don't know how to make this independent, of cause you would have to replace it with a function
[01:58:55] <slaad> #define strlcpy(dest, data, length) strncpy(dest, data, length - 1)
[01:58:58] <fyysik> Dr_Evil - any change in multiaudio add on recently?
[02:00:40] <@Dr_Evil> not that I know
[02:01:14] <@Dr_Evil> #define strlcpy(dest, data, length) do { strncpy(dest, data, length - 1); dest[length-1] = 0; } while (0)
[02:01:23] <@Dr_Evil> terminating with 0 is important
[02:01:32] *** MikeW has joined #haiku
[02:02:18] <fyysik> Dr_Evil - any reason to use snooze() in drivers, not spin() ?
[02:02:19] <[Beta]> problem in returning the new var size ?
[02:02:50] <ShackaN> Dr_Evil, a question
[02:03:12] <ShackaN> it's not the first time I see a do { ... } while(0);
[02:03:26] <ShackaN> although that a total nonsense for me
[02:03:46] <ShackaN> why don't you rather use a simple pair of braces ?
[02:03:52] <ShackaN> { ... }
[02:04:03] <@Dr_Evil> fyysik snooze releases cpu to other tasks, and should be the normal use. however, if you ned very short timing, and want to avoid sheduling, or inside an interrupt handler, where it is not allowed, you need to use spin
[02:04:23] <fyysik> thanks Dr_Evil
[02:04:25] <@Dr_Evil> ShackaN for C compiler backward compatibliity
[02:04:40] <@Dr_Evil> however, with gcc the simple way might be ok
[02:04:50] <ShackaN> ok
[02:07:03] <ShackaN> Dr_Evil, can Haiku be compliled with gcc 3.3.5 ?
[02:07:16] <ShackaN> that's the only compiler I have atm
[02:07:18] <@Dr_Evil> dont know
[02:07:23] <ShackaN> ok
[02:09:17] <@Dr_Evil> chatting disturbs my development
[02:11:38] <[Beta]> ooh, paypal :>
[02:16:40] <[Beta]> 2 01:16:40 <[Beta]>	2 Jan. 2005 	 Payment To Haiku, Inc
[02:18:57] *** ablyss has joined #haiku
[02:21:12] <@Dr_Evil> you can also donate to me personally (will not be forwarded to Haiku)
[02:22:16] <[Beta]> lol
[02:22:32] <[Beta]> I dont give money to evil people :))
[02:22:46] <@Dr_Evil> just use my email marcus at overhagen dot de when donating with paypal
[02:24:41] <[Beta]> hey, it's christmas, i'm poor :p
[02:24:58] <[Beta]> maybe next month :)
[02:27:05] <@Dr_Evil> hmm, anyone?
[02:29:30] <fyysik> ?
[02:32:09] *** Striver has quit IRC
[02:33:44] <slaad> While people are donating... perhaps throw some the IM Kit's way ;)
[02:34:08] <ShackaN> hmmm
[02:34:14] <ShackaN> gr?????d
[02:34:25] <slaad> ?
[02:34:29] <slaad> You appear to be b0rked.
[02:36:04] <[Beta]> I still reckon irc is in the realm of im
[02:37:54] <slaad> As in, you want IRC support in the IM Kit?
[02:38:56] <[Beta]> might as well go the whole hog.
[02:39:34] <@voidref> that would require quite a different im client
[02:39:56] <slaad> Yeah... personally, I don't want a bar of it.
[02:40:13] <slaad> I'm yet to see a good implementation of IRC in an IM client. And ... well.. Vision kicks ass :)
[02:40:15] <[Beta]> i'm just on about an API
[02:40:48] <[Beta]> why have a Be way for IM, and a per-app way for IRC ?
[02:41:02] <@voidref> idref>	i had tried to get the yak to implement IM in VIsion at one point, and he said that it was quite a different beast, from a code and logic flow standpoint.
[02:41:33] <@voidref> slaad, actually, I remember trillian being pretty nice looking
[02:41:42] <@voidref> although lots of times it didn't work quite right
[02:41:47] *** ConneX- has joined #haiku
[02:41:52] <@voidref> due to buggy code, I would imagine
[02:41:53] * fyysik thinks it'll be nice tho have also cmpci multiaudio version
[02:42:11] <fyysik> my another sound is built-in cmpci
[02:42:12] <slaad> Yeah, it is... because IM is mostly one to one. Where as IRC is one to many.
[02:42:37] <[Beta]> not if you look at it like that
[02:42:38] <slaad> Trillian isn't too bad. But it's more of an IRC client than IRC in an IM client. If you get the difference.
[02:49:38] <[Beta]> nn
[02:50:27] <slaad> Are there any open source muscle apps about?
[02:52:30] <ShackaN> miranda-im, maybe, nevere used it tough
[02:52:35] <ShackaN> *never
[02:56:54] *** Begasus has quit IRC
[02:59:16] *** voidref has quit IRC
[03:01:01] *** voidref has joined #haiku
[03:01:02] *** ChanServ sets mode: +o voidref
[03:08:03] *** DaaT has quit IRC
[03:08:26] *** fyysik has quit IRC
[03:21:02] *** Dr_Evil has quit IRC
[03:23:53] *** Potn has quit IRC
[03:24:26] *** Potn has joined #haiku
[03:25:34] *** ablyss has quit IRC
[03:27:52] *** Potn has quit IRC
[03:28:32] *** Potn has joined #haiku
[03:32:29] *** LupusMichaelis has joined #haiku
[03:43:18] *** voidref has quit IRC
[03:43:43] *** Potn has quit IRC
[03:44:20] *** Potn has joined #haiku
[03:53:43] *** relix has joined #haiku
[03:55:14] <relix> we study, study
[03:55:21] <relix> what the hell is entropy?
[03:55:25] <relix> thermodynamics
[03:58:11] *** Dr3w has joined #haiku
[04:00:28] *** relix has quit IRC
[04:04:51] *** inseculous has quit IRC
[04:23:47] *** ShackaN has quit IRC
[04:29:04] *** MikeW has quit IRC
[04:38:58] *** mmadia has joined #haiku
[05:34:13] *** Greyhawk_Dragon has joined #haiku
[05:42:48] *** brennanos has joined #haiku
[05:43:15] <brennanos> hi
[05:44:49] *** Dr3w has quit IRC
[05:59:43] *** MikeW has joined #haiku
[06:10:15] *** brennanos has left #haiku
[06:31:10] *** BetaMax has joined #haiku
[06:38:50] *** voidref has joined #haiku
[06:38:50] *** ChanServ sets mode: +o voidref
[07:01:23] *** BetaMax has quit IRC
[07:16:07] *** MikeW has quit IRC
[07:28:23] <mmadia> are there any AMD cpus that beos won't work on?   eg, opteron or athlon 64
[07:43:09] <nPHYN1T3> no
[07:43:33] <nPHYN1T3> BeOS runs on AMD 64 and Opterons fine
[07:43:58] <mmadia> thanks nPHYN1T3.
[07:44:05] <nPHYN1T3> yup
[07:45:15] <nPHYN1T3> I have booted 4 4.5 5 and 5.1 on Dual Opterons
[07:45:41] * mmadia drools
[07:45:44] <nPHYN1T3> they run in 32bit mode & Be is none the wiser
[07:46:50] <nPHYN1T3> I really would like to have seen Haiku focus on 64 bit chips and massive memory
[07:47:23] <nPHYN1T3> I know you see a BIG differance between 32bit & 64bit OS
[07:47:29] <nPHYN1T3> difference
[07:48:09] <mmadia> there's always R2+
[07:48:10] <nPHYN1T3> Windows XP 32 running premier is much slower then WinXP 64 running premier
[07:48:15] <nPHYN1T3> ya
[07:49:17] <mmadia> although i dont know much, i'd imagine switching the codebase to 64bit would be a pita esp. as it's still an in-the-minority technology.
[07:49:38] <nPHYN1T3> ya but thats why I say I'd like to have seen that be the focus
[07:49:45] <nPHYN1T3> 64bit native code base
[07:49:48] <nPHYN1T3> look forward
[07:50:16] <nPHYN1T3> it will be a while till the OS is released as R1 might as well make it better adapted to the hardware being used around that time
[07:50:17] <mmadia> hard to code on 32-bit cpus probably :-)
[07:50:25] <nPHYN1T3> it wont be long till the 64bits saturate
[07:50:32] <nPHYN1T3> ya I know
[07:50:39] <nPHYN1T3> gotta have 64's for the devs :)
[07:51:16] <nPHYN1T3> I'd rather see support for 64bit AMD & 64bit IBM chips though
[07:51:46] <mmadia> power5 ?
[07:51:56] <nPHYN1T3> ya
[07:52:10] <nPHYN1T3> power 5 and 970's
[07:52:28] <nPHYN1T3> chips where we can see 64gigs -o- ram :)
[07:53:07] <nPHYN1T3> I think it was said R1 will only be able to deal with 2 gigs of ram - thats dissapointing
[07:54:37] <mmadia> if settling for 512mb would help R1 or even PR1 get out the door, i'd be happy.
[07:54:58] <nPHYN1T3> what do you mean?
[07:55:14] <nPHYN1T3> if R1 can only use 512mb I won't be able to run it on any of my systems lol
[07:55:52] <mmadia> eer,, in a way similar to the ram-limiting patch.
[07:56:09] <nPHYN1T3> I dunno why the hell they capped it at 512
[07:56:14] <nPHYN1T3> they should have capped at 768
[07:56:24] <mmadia> i'm more than content dealing with limitations of a pre-R2 system than dealing with not having it altogether.
[07:57:07] <nPHYN1T3> I'd want a min of 768mb supported
[07:57:12] <nPHYN1T3> at least R5 can do a gig
[07:57:41] <mmadia> IMHO, having *something* that the masses can easily install and play with is a monumental milestone.
[07:58:26] <mmadia> ...it should bring new and old users back to the community, and hopefully more developers.
[07:58:32] <nPHYN1T3> ya but most the masses that tinker with Alt OS's have serious machines for doing lots of stuff & hence more then 512mb of ram
[07:59:01] <mmadia> i'm using 512mb for arguments sake :-)
[07:59:01] <nPHYN1T3> so unless the kaiku kernel caps the ram like the patch right off lots of people won't be able to easilly install & play with it
[07:59:25] <nPHYN1T3> this machine has 768mb & my others are at a gig
[07:59:31] <mmadia> agreed.
[08:00:28] *** kr1stof has quit IRC
[08:02:30] <nPHYN1T3> I tell you what though
[08:02:45] <nPHYN1T3> if I win the lottery I'll be hiring devs like mad to get shit moving
[08:02:52] <mmadia> :-D
[08:02:56] <nPHYN1T3> cuz I'm sick of limits & waiting :)
[08:03:28] <mmadia> ia>	i considered selling kidneys but i couldn't get anyone drunk enough to make them pass out in a tub of ice.
[08:04:06] <nPHYN1T3> heh
[08:09:38] * mmadia would love for someone to create a definitive list or .pkg of installable (and relatively stable) Haiku components
[08:10:43] <nPHYN1T3> I tried starting to add haiku stuff to my system but it broke lots
[08:12:04] <mmadia> same here.
[08:28:16] *** Koki has quit IRC
[09:07:08] *** sl44d has joined #haiku
[09:13:58] *** SeaForth has joined #haiku
[09:24:56] *** slaad has quit IRC
[09:24:56] *** sl44d is now known as slaad
[09:31:30] *** Parker has joined #haiku
[09:35:51] *** Greyhawk_Dragon has quit IRC
[09:36:09] *** Parker has quit IRC
[09:53:01] *** kr1stof has joined #haiku
[10:11:41] *** SeaForth has quit IRC
[10:25:01] *** FastJack has quit IRC
[10:28:00] *** Greyhawk_Dragon has joined #haiku
[10:29:28] *** zortness has quit IRC
[10:34:23] *** FastJack has joined #haiku
[10:53:50] *** tqh has joined #haiku
[11:32:31] *** Dr3w has joined #haiku
[11:45:12] *** Potn has quit IRC
[11:51:44] *** Potn has joined #haiku
[12:06:13] *** Master199 has joined #haiku
[12:10:41] *** nPHYN1T3 has quit IRC
[12:11:22] *** fyysik has joined #haiku
[12:12:15] *** khorben has joined #haiku
[12:39:46] *** sl44d has joined #haiku
[12:42:33] <mmadia> did i miss any supported filesystems?  RAMFS, FTP-FS, BigFS,  BFS, FAT, NTFS, HFS, ext2, ReiserFS, AtheOS, ISO9660, UDF, cdda, and cifs.
[12:43:55] <mmadia> that are supported in BeOS.
[12:43:56] <@voidref> DOS?
[12:44:28] <@voidref> ftp-fs isn't exactly supported well
[12:45:02] <@voidref> isn't ext3 supported by ext2 support, at least for reads?
[12:45:39] <mmadia> idunno.  haven't used ext3.
[12:46:13] <@voidref> there's somethign called ofs in my file_systems kernel addons dir
[12:46:18] <@voidref> and uspacefs
[12:46:39] <@voidref> ah, FAT is DOS, sorry I missed that
[12:46:51] <sl44d> DOS is just FAT12.
[12:46:54] <sl44d> Silly FS that is.
[12:47:06] <sl44d> "Look at us! We use 1 1/2 bytes per entry. Woo!"
[12:47:16] <@voidref> 8.3 baby!
[12:47:36] <sl44d> You can still do VFAT on it :)
[12:47:38] <mmadia> 8letters.wow
[12:47:49] * sl44d pets his FAT drivers for 12 bit micros.
[12:48:02] <@voidref> micros~1?
[12:48:09] <sl44d> Haha
[12:48:17] <sl44d> Not quiet :)
[12:48:18] *** slaad has quit IRC
[12:48:18] *** sl44d is now known as slaad
[12:48:26] <@voidref> it's loud?
[12:48:38] <mmadia> windo~1.sux   :)
[12:48:39] <slaad> Curses.
[12:48:42] <@voidref> what is the difference between l33t spelled slaad and not?
[12:48:42] <slaad> It's 11, baby!
[12:49:04] <slaad> One is my router sucking absolute wang. One is not.
[12:49:48] *** fyysik has quit IRC
[12:50:09] <@voidref> ah
[12:50:25] <@voidref> get new firmware for that router.
[12:50:30] <slaad> Doesn't help.
[12:50:35] <slaad> The router just hardlocks.
[12:50:38] <slaad> Needs a power cycle.
[12:51:00] <slaad> So word of advice; Use IP over carrier pigeon before buying a D-Link router.
[12:57:41] <@voidref> well, even my old linksys would need a reset every few months.
[12:58:00] <@voidref> idref>	i am hoping the new one performs better
[12:58:47] <slaad> We're talking every few hours - days.
[12:59:05] <slaad> Total hardlock. Doesn't even do normal hub / switch activity.
[12:59:11] <slaad> Or respond to any packets :(
[12:59:25] <[Beta]> erm, I dont have a problem with my D-Link router
[12:59:38] <[Beta]> though its got the worst html interface i've ever seen.
[13:00:27] <[Beta]> what model, slaad?
[13:01:18] <slaad> Err... 5-0-wang. Or something. 508? Maybe? I don't know.
[13:01:32] <slaad> Something like that.
[13:02:26] <[Beta]> i've got a 504t, adsl router w/ shitty firewall
[13:02:38] *** gipfex has joined #haiku
[13:02:52] <[Beta]> least all its software is open source, so I can replace the minging interface.
[13:03:01] <slaad> It is?
[13:03:10] <[Beta]> download it from the website, sure.
[13:03:14] <slaad> Interesting.
[13:03:34] <slaad> I'm thinking I should just use it in bridge mode. And have a computer do all the NAT.
[13:04:46] <[Beta]> kindof a waste of getting a router then ?
[13:06:06] <slaad> It was free.
[13:06:12] <slaad> And besides, have you ever tried buying a DSL modem?
[13:06:15] <[Beta]> oh, bonus.
[13:06:29] <[Beta]> yeah, thats it ^^
[13:06:32] <slaad> Not really, if I didn't get it, I might have bought a good one :)
[13:09:58] <slaad> Ung... man... writing a new protocol addon is going to be a pain in the ass with this slow startup / shutdown
[13:10:57] <w-ber> http://www.digitalmars.com/d/intro.html
[13:12:23] *** illissius- has joined #haiku
[13:12:46] <mmadia> slaad which protocol?
[13:13:01] <slaad> The Proxy protocol.
[13:13:17] <slaad> Which connects to an instance of im_proxy_server. Which is an IM client.
[13:13:25] <slaad> And relays messages between them.
[13:13:44] <mmadia> for MYOB?
[13:13:50] <slaad> I'm not sure if it'll actually work yet.. but.. it won't take long to write.
[13:14:03] <slaad> For MYOB? No... what's with him?
[13:14:32] <mmadia> he's behind a "proxy server from hell"   cant connect to irc, beshare, or darn near anything.
[13:14:40] <slaad> Oh well, sucks to be him :)
[13:14:59] <slaad> Oh, yeah, I remember now. He seems to be interested in the IMWebClient.
[13:15:16] <mmadia> as are other people :)
[13:15:39] *** markjan has joined #haiku
[13:15:56] <slaad> They haven't commented in my blog!
[13:15:58] * slaad runs off crying.
[13:20:34] *** illissius has quit IRC
[13:22:25] *** gipfex has left #haiku
[13:23:39] *** [Beta] has quit IRC
[19:34:14] *** [Beta] has joined #haiku
[19:40:22] *** Potn has joined #haiku
[19:47:30] *** UndeadYak has joined #haiku
[19:47:31] *** Fanskapet has quit IRC
[19:47:40] *** Fanskapet has joined #haiku
[19:47:42] *** UndeadYak has quit IRC
[19:49:43] *** sys2 has quit IRC
[21:12:06] *** [Beta] has joined #haiku
[21:12:25] *** tic has quit IRC
[21:14:23] *** fyysik has joined #haiku
[21:21:58] *** ShackaN has joined #haiku
[21:26:24] *** z3r0_one has quit IRC
[21:32:59] *** Fanskapet has joined #haiku
[21:48:27] *** LupusMichaelis has quit IRC
[21:49:58] *** brennanos has joined #haiku
[21:51:50] *** Potn has quit IRC
[21:52:29] *** Potn has joined #haiku
[21:58:18] *** TLF has joined #haiku
[22:06:12] *** thies is now known as Apu
[22:11:53] *** Leo42 has joined #haiku
[22:12:11] <Leo42> hi
[22:12:50] *** tic has joined #haiku
[22:16:07] <Methe> lo
[22:17:46] *** LupusMichaelis has joined #haiku
[22:23:19] <reffie> lo
[22:26:57] *** Konrad has quit IRC
[22:29:31] *** TLF has quit IRC
[22:30:14] <fyysik> tqh - still stick with build?
[22:30:18] <fyysik> stuck
[22:30:56] <Leo42> what's new on haiku ?
[22:32:13] *** flameshadow has joined #haiku
[22:34:11] <flameshadow> How does BeFS track which inodes are directories and which ones are files?
[22:34:56] <AnEvilYak> don't remember for certain offhand, lemme get my BFS book
[22:36:06] <flameshadow> I see that INODE_IN_USE is tagged as 0x00000001, and ATTR_INODE is 0x00000004 .. but it doesn't mention what is 0x0000002 ..
[22:36:31] <flameshadow> And I'm wondering if that's how they tell what's a directory
[22:37:37] *** Potn has quit IRC
[22:38:15] *** Potn has joined #haiku
[22:41:17] <AnEvilYak> just uses the S_IFDIR flag.
[22:41:20] <AnEvilYak> in the mode bits.
[22:41:22] <AnEvilYak> nothing beyond that.
[22:41:39] <flameshadow> Do you know what the value of S_IFDIR is?
[22:41:49] <AnEvilYak> that's POSIX..
[22:41:56] <AnEvilYak> osix/sys/stat.h:#define     S_IFDIR                     00000040000 /* directory */
[22:42:02] <flameshadow> Awesome.. thanks.
[22:42:06] <flameshadow> One other question, if you don't mind..
[22:42:10] <AnEvilYak> shoot.
[22:42:30] <flameshadow> It says that the maximum block extent is 65536 because the len field in a blockRun is 16-bit...
[22:42:35] <tqh> fyysik Dunno, think it might be one of my patches from AVIARY. I'm trying to see which.
[22:42:36] <AnEvilYak> right.
[22:42:38] <flameshadow> But, the largest value you can stick in there is 65535..
[22:42:44] <flameshadow> So, I see three possibilities
[22:42:52] <flameshadow> 65535 is the largest block run you can have
[22:42:59] <flameshadow> 02 21:42:59 <flameshadow>	0 represents a block run of 65536
[22:43:03] <flameshadow> or everything is shifted by one
[22:43:20] <flameshadow> (0 is 1 block run, 1 is 2 blocks, etc)
[22:43:23] <AnEvilYak> 65535 would be the correct answer I believe.
[22:44:11] <flameshadow> Again, many thanks.
[22:44:51] <flameshadow> Hm
[22:45:05] <flameshadow> NO_TRANSACTION seems to share the same value as S_IFDIR
[22:45:05] <@Dr_Evil> I could look it up in the BFS book if required
[22:46:11] <tqh> fyysik I did test my aviary build with extentions while building and it worked great, I'm now using swedish locales and have a weatherforcast for two days next to the URL-bar.
[22:46:25] <AnEvilYak> flameshadow: that's used in a different field
[22:46:27] <AnEvilYak> flameshadow: no conflict.
[22:46:55] <flameshadow> Awesome. You guys rock :)
[22:47:20] *** slaad has joined #haiku
[22:49:47] <@Dr_Evil> The BFS book says that a block run can map 65536 blocks, but it doesn't tell how
[22:52:10] <flameshadow> Yes.. hence my confusion
[22:52:35] <frankps> Does the bookmark sync extension work under Zeta/BeOS?
[22:52:42] <@Dr_Evil> however, this book is full of errors
[22:52:44] <frankps> The one for Firefox
[22:53:07] <@Dr_Evil> flameshadow you should check the openbfs source
[22:53:25] <flameshadow> I'm trying to do a clean-room re-implementation
[22:53:43] <tic> why?
[22:53:52] <tic> isn't it a lot easier to use the Haiku sources?
[22:53:58] <flameshadow> I'm not sure that it is.
[22:54:05] <tic> okay...
[22:54:12] <tic> for which OS are you writing it?
[22:54:15] <flameshadow> UbixOS
[22:54:15] *** Begasus has joined #haiku
[22:54:22] <tic> an 02 21:54:22 <tic>	a un*x derivate?
[22:54:37] <flameshadow> Little bit of everything.
[22:54:41] <tic> okay.
[22:54:52] <flameshadow> GUI and FS based on BeOS
[22:55:04] <tic> does it run Tracker?
[22:55:11] <flameshadow> I'm not quite that far
[22:55:19] <tic> what kernel are you using?
[22:55:23] <flameshadow> We wrote our own
[22:55:25] <@Dr_Evil> flameshadow so why don't you extend it and switch to 32 bit block_run length, etc?
[22:55:27] <tic> a-ha.
[22:55:36] <flameshadow> Dr_Evil: that won't fit into an 8-byte field
[22:55:49] <flameshadow> And I'm not picky, I just want to have a close implementation, not a 100% perfect one
[22:56:06] <@Dr_Evil> BFS was designed when it was cheap to read a few small blocks scattered over the disk. today it's cheaper to read a chunk of a few MBs
[22:56:23] <tic> Hm..
[22:57:35] <@Dr_Evil> linear transfer speed has increased enormously
[22:57:56] <tic> Dr_Evil, does OpenBFS exploit this?
[22:58:11] <Methe> tic: u mean HaikuFS ? :x
[22:58:27] <tic> heh.
[22:58:31] *** Potn_ has joined #haiku
[22:59:23] <tic> hey #haiku
[22:59:23] <@Dr_Evil> tic well cache and bfs allocation policies can be tuned to account fro that
[22:59:25] <tic> err, Potn
[22:59:30] <tic> Dr_Evil, that's good.
[22:59:43] <@Dr_Evil> but when writing a new FS, I would make certain things different
[22:59:49] *** Potn has quit IRC
[22:59:49] <tic> okay.
[23:00:00] <@Dr_Evil> Methe there is no HaikuFS
[23:00:09] *** tqh has quit IRC
[23:00:25] <Methe> Dr_Evil: ah ? OBFS kept its name ?
[23:00:29] <AnEvilYak> from looking at the BFS src, 0-65535 would appear to be the case.
[23:00:41] <flameshadow> Thanks, AnEvilYak.
[23:00:57] *** illissius- is now known as illissius
[23:01:25] <AnEvilYak> least I don't see any special casing or whatnot.
[23:01:30] <@Dr_Evil> according to the book, the raw read speed of BFS was 5.8 MB/sec when it was written
[23:01:56] <@Dr_Evil> Methe yes, so we can do a native haikuFS if we want :)
[23:02:16] <Methe> Dr_Evil: well played :)
[23:02:18] <@Dr_Evil> raw read speed is 38 to 65  MB/sec on modern harddisk
[23:02:32] *** Potn_ has quit IRC
[23:02:59] <@Dr_Evil> so with one seek, you can today read 10 times the amount of data
[23:03:03] *** Potn_ has joined #haiku
[23:03:23] <@Dr_Evil> however, seek speed hasn't impoved much from 14 ms to 8 ms or so
[23:06:23] <AnEvilYak> http://www.bash.org/?432558
[23:06:23] <AnEvilYak> lmao
[23:06:50] <AnEvilYak> Dr_Evil: yeah, which is why B+ trees are as important as ever.
[23:07:43] <@Dr_Evil> yes, but allocation policies and block run length should be tuned
[23:08:09] <@Dr_Evil> considering a 400 GB harddisk, and storing a 15 GB DV movie on it
[23:08:10] <AnEvilYak> allocation policy is tricky nowadays.
[23:08:23] <AnEvilYak> because the disk microcontroller abstracts the physical disk structure nowadays
[23:08:30] <flameshadow> But do you really need to read in 65 contiguous meg at one clip?
[23:08:33] <AnEvilYak> so you can't rely on physical placement via block indexes or w hatnot any more.
[23:08:35] <@Dr_Evil> the 45 minutes that I recorded on a friends marriage are 10.7 GB DV video
[23:08:43] <AnEvilYak> right
[23:08:57] <AnEvilYak> XFS is nice that way.
[23:09:02] <@Dr_Evil> and this is stored in 1 kb blocks on a standard BFS disk
[23:09:03] <AnEvilYak> can map huge files ridiculously easily.
[23:09:14] <@Dr_Evil> even a block rn can only map 65535 of them
[23:09:15] <AnEvilYak> iirc an XFS extent map can do a few million blocks with one run.
[23:09:26] <tic> 6x200G w/o VAT is US$620. Would make a nice 1TB RAID5.
[23:09:32] <AnEvilYak> well, BFS could support huge files without too much issue with triple indirect implemented.
[23:09:35] <AnEvilYak> which iirc OBFS does.
[23:09:42] * Dr_Evil recently bought 4* 200GB
[23:09:46] <tic> Dr_Evil, raid?
[23:09:58] <AnEvilYak> iirc dbg said implementing triple indir would push max file size to around 18TB
[23:10:01] <@Dr_Evil> used it with windows. RAID 5 was terribly slow, 6 MB/sec write performance
[23:10:14] <flameshadow> You could also up the amount of blocks allocated in the double-indirect
[23:10:16] <tic> AnEvilYak, if you used large inode sizes, yeah.  on 1024 you're limited to a lot less than that.
[23:10:16] <@Dr_Evil> now using 2 as mirror, and 2 as stripeset
[23:10:24] <AnEvilYak> tic: *nod*
[23:10:26] <tic> Dr_Evil, okay.
[23:10:36] <AnEvilYak> tic: in the current BeOS kernel you have to be careful with large block size though.
[23:10:45] <AnEvilYak> tic: cache has some calculations in it that assume 1KB block size
[23:10:56] <@Dr_Evil> tic that gives about 24 MB/sec write on the stripeset, and 16 MB/sec on the mirror
[23:10:56] <AnEvilYak> tic: and if you use a different size it screws up and ends up allocating a huge amount for cache.
[23:10:56] <tic> Dr_Evil, I was thinking 5 disks for storage, 1 for parity => 1TB.
[23:11:01] <tic> Dr_Evil, yah.
[23:11:11] <tic> AnEvilYak, bleh.  So fix that! :P
[23:11:21] <@Dr_Evil> however, the PCI SATA controller, combined with nforce2 chipset sucks
[23:11:29] <AnEvilYak> note that that was with the R5 kernel anyhow.
[23:11:30] <tic> that's nasty.
[23:11:34] *** Potn_ has quit IRC
[23:11:36] <@Dr_Evil> raw disk perfromance was only about 30 MB, should have been 60
[23:11:38] <tic> AnEvilYak, okay.  might not be the case with R5.1+ ?
[23:11:45] <AnEvilYak> tic: dunno, me not kernel guy
[23:11:46] <tic> Dr_Evil, strange indeed.
[23:11:51] <tic> AnEvilYak, okay. who be kernel guy?
[23:12:00] <AnEvilYak> tic: I generally only know what I've been told by various engineers in the past.
[23:12:04] <tic> AnEvilYak, gotcha
[23:12:05] <AnEvilYak> tic: and/or by mmu_man ;p
[23:12:14] <tic> AnEvilYak, hehe.
[23:12:17] <tic> he's quite the hacker.
[23:12:20] <AnEvilYak> yeah
[23:12:24] <AnEvilYak> if he really does KDL pong I'll be scared.
[23:12:32] <@Dr_Evil> tic no a sofware raid 5 is slow
[23:12:33] <tic> how the heck would that work?
[23:12:40] <AnEvilYak> tic: don't ask me, that's why I'm scared.
[23:12:45] <tic> AnEvilYak, haha :P
[23:12:45] <AnEvilYak> tic: I know he was considering it at some point ;p
[23:12:52] <AnEvilYak> tic: but hey, he did KDL hangman...
[23:12:54] <tic> AnEvilYak,  heh.  I like jbq's BWindow-pong.
[23:12:54] <@Dr_Evil> tic I read a perfomcenace test of a computer magazne that uses raptor drives, and they got write speed of about 9 MB/sec
[23:13:01] <AnEvilYak> tic: that was awesome
[23:13:03] <tic> Dr_Evil, jikes!  that's really slow.
[23:13:16] <tic> AnEvilYak, yeah.  And you could do a lot better in R5.1+ w/ ClipWinToPic
[23:13:21] <AnEvilYak> tic: *nod*
[23:13:31] <tic> Dr_Evil, but I've read that the IDE RAID cards are really just software cards anyway.
[23:13:37] <tic> Dr_Evil, RAID5, that is.
[23:13:39] <@Dr_Evil> yes
[23:13:55] <tic> so a real raid card would cost quite a lot, right?
[23:14:01] <@Dr_Evil> thats why I used windows software raid and a normal controller
[23:14:06] <tic> mhm
[23:14:22] <@Dr_Evil> tic  I was a little disappinted by the speed result after spending money on 4 200GB sata harddisks
[23:14:27] <tic> Dr_Evil, :/
[23:14:27] <AnEvilYak> I'm happy with my 39160 ;p
[23:14:29] <tic> I can tell.
[23:14:34] <AnEvilYak> not RAID but whatever.
[23:14:37] <@Dr_Evil> but at least I borrowed the Promise SATa controller
[23:15:30] <tic> speaking of IDE controllers, I bought a promise ata100tx2 not long ago. going against the stream. :)
[23:16:27] <tic> so now I have three IDE controllers in my 'puter
[23:16:34] <@Dr_Evil> this is a Promise TX4 SATA
[23:16:34] <tic> i440bx, hpt366 & promise
[23:16:34] *** illissius is now known as illissius[sleep]
[23:16:42] <tic> mhm.  promise makes good cards, it seems.
[23:16:51] * tic is leaning towards a card w/ a built-in SATA drive.
[23:16:55] <tic> err, promise controller
[23:16:58] * tic can't type
[23:16:59] <@Dr_Evil> yes, my motherboard sucks
[23:17:01] <tic> s/card/mobo
[23:17:10] <@Dr_Evil> can even set PCI latency timer default in BIOS
[23:17:30] <tic> Dr_Evil, get an AOpen i855-something and a Pentium M 735 (1.7 GHz), which you then can either a) run silently, or b) oveclock like mad. :)
[23:17:53] <AnEvilYak> bbiab
[23:18:00] *** AnEvilYak has quit IRC
[23:19:40] <@Dr_Evil> flameshadow bfs.h:// break compatibility and take a zero length for 65536).
[23:20:58] *** frankps has left #haiku
[23:21:05] <@Dr_Evil> BlockAllocator.cpp:     // a) since a length of zero doesn't have any sense, take that for 65536 -
[23:23:18] <flameshadow> hm. Alright.
[23:23:44] *** Apu has quit IRC
[23:24:24] *** brennanos has left #haiku
[23:27:10] *** thies has joined #haiku
[23:27:11] <flameshadow> So, in the original it was limited to 65535, but openBeOS^WHaiku allows for that extra block.
[23:28:16] <@Dr_Evil> yes
[23:29:06] <@Dr_Evil> I think we just found out why openbfs can destroy bfs partitions and make them unreadable for R5
[23:29:51] <[Beta]> ooh, historical.
[23:29:53] <flameshadow> Does your version of BeFS extend previous block runs if they are contiguous to the last block run (i.e. when you're extending the file)?
[23:30:23] <@Dr_Evil> dont know
[23:31:05] <flameshadow> If you do, using 0 as 65536 would seem to complicate the calculation of which was the last block run, since you can't just scan for the first 0 length blockrun, you'd have to add them all up
[23:32:19] <@Dr_Evil> talk to axeld, he should be here in about 24 hours and 30 minutes
[23:32:24] <@Dr_Evil> he wrote most of openbfs
[23:33:24] <flameshadow> I can hang about for a bit.
[23:33:56] <tic> Dr_Evil, what's happening tomorrow at midnight?
[23:36:58] <@Dr_Evil> meeting time
[23:37:14] <tic> nice.  what are you guys going to talk about?
[23:37:24] <@Dr_Evil> every monday, on 0:00 (local time germany)
[23:37:36] <tic> yeah.
[23:38:07] <@Dr_Evil> probably talking about this and that and the new webhosting
[23:38:31] <tic> *nods*
[23:38:42] <tic> anything else in particular?
[23:39:12] <@Dr_Evil> well, status, progress, plans, etc
[23:39:24] <tic> mhm. same old, same old?
[23:42:41] <mumu25> in #haiku, or in the special #haiku-team-leader-in-scene-bar
[23:43:46] <tic> scummbar
[23:44:05] <[Beta]> prolly the latter :/
[23:44:10] <@Dr_Evil> scene bar
[23:57:48] *** thaflo has joined #haiku





top