| 00:02.20 | pb_ | night. |
| 00:18.05 | *** join/#gpe dc___ (~dc@modem-200-53-60-62.vip.uk.com) |
| 00:24.29 | dc___ | sleeeeeeeeep |
| 00:52.41 | *** join/#gpe jg_ (~jg@h005018015b26.ne.client2.attbi.com) |
| 01:29.27 | *** join/#gpe NonToxic (~joshua@got.owned.by-the.us) |
| 01:29.39 | NonToxic | would anyone know how to add an icon in gpe? |
| 01:30.03 | PaxAnima | add to mbdesktop, or add to the menu? (or both?) |
| 01:30.10 | NonToxic | both |
| 01:31.35 | PaxAnima | create a .desktop file and put it in /usr/share/applications |
| 01:31.39 | NonToxic | ok |
| 01:33.11 | NonToxic | k.... |
| 01:33.13 | NonToxic | didn't work... |
| 01:33.27 | PaxAnima | you must log in and out, I think |
| 01:34.15 | PaxAnima | may out and in is better... ;) |
| 01:34.46 | NonToxic | still not working |
| 01:34.47 | NonToxic | hold on |
| 01:35.22 | NonToxic | ah |
| 01:35.23 | NonToxic | it needs an icon |
| 01:40.19 | NonToxic | w00t |
| 01:40.25 | *** join/#gpe paul ([+i+LU9grs@hibernia.jakma.org) |
| 01:40.26 | NonToxic | sfcave has an icon, and works! |
| 01:40.46 | NonToxic | will be tarring up shortly. |
| 01:41.40 | PaxAnima | cool |
| 01:43.07 | *** join/#gpe flypiper (~deb-user@12-222-141-210.client.insightBB.com) |
| 01:43.37 | NonToxic | tarring now. |
| 01:44.16 | PaxAnima | i.e. creating an ipkg? |
| 01:44.46 | NonToxic | I haven't a clue how to create an ipkg :) |
| 01:45.16 | PaxAnima | ah.. pretty straight forward, really... |
| 01:45.21 | NonToxic | http://www.joshuawise.com/~joshua/sfcave.tar.gz - requires: libsdl >= 1.2-jw |
| 01:45.46 | NonToxic | http://www.joshuawise.com/~joshua/sdl.tar.gz - provides: libsdl 1.2-jw |
| 01:46.03 | NonToxic | I've tried it before and not had success |
| 01:48.34 | PaxAnima | have a look at ~paxanima/pscalc on the ipaq cluster... it's really not that difficult |
| 01:49.25 | PaxAnima | I could try and make an ipkg for you |
| 01:50.03 | NonToxic | hrm... |
| 01:50.09 | NonToxic | if you could, that would be great |
| 01:50.37 | PaxAnima | wait 2 secs (or more) |
| 01:50.41 | NonToxic | l |
| 01:50.44 | NonToxic | k |
| 01:55.02 | PaxAnima | created the ipkg.. I'll just test |
| 01:55.03 | PaxAnima | it |
| 01:55.09 | NonToxic | k |
| 02:00.43 | PaxAnima | hmm |
| 02:01.13 | PaxAnima | what's special with your SDL lib? |
| 02:04.32 | NonToxic | PaxAnima: it was compiled to use x11, and it has various other libs that are needed. |
| 02:04.51 | PaxAnima | ah |
| 02:08.40 | PaxAnima | how do I start a game? |
| 02:09.01 | NonToxic | open xkbd and hit space |
| 02:10.47 | PaxAnima | yay! it works! should hack on the code and make it the enter key rather than space, so it's possible to operate witht the keypad |
| 02:11.24 | NonToxic | yah |
| 02:11.34 | NonToxic | its a little slow |
| 02:11.35 | NonToxic | oh well |
| 02:12.11 | PaxAnima | worked fine here |
| 02:12.25 | NonToxic | yeah, I mean, it just sucks cycles |
| 02:12.33 | PaxAnima | ah |
| 02:13.11 | PaxAnima | got the ipkgs at http://handhelds.org/~paxanima/files/ |
| 02:13.29 | NonToxic | k. |
| 02:13.30 | NonToxic | thanks. |
| 02:14.48 | PaxAnima | should take a closer look at the ipkgs, with regards to conflicts with the SDL packages already in the feeds, but if you uninstall libsdl, then install those ipkgs, it should work fine |
| 02:15.22 | NonToxic | yeah |
| 02:15.42 | *** join/#gpe walters (walters@verbum.org) |
| 02:15.46 | PaxAnima | they are not uploaded (actually, I'm the maintainer, acording to the control file;) |
| 02:17.34 | NonToxic | hehe |
| 02:17.58 | NonToxic | we now have the advantage over opie! |
| 02:18.02 | NonToxic | speed! looks! AND games! |
| 02:18.02 | PaxAnima | NonToxic: the ipkg control files can be found at ~paxanima/sfcave-ipkg and ~paxanima/sdl-jw-ipkg if you want to change that and create new ones |
| 02:18.04 | NonToxic | :) |
| 02:18.06 | PaxAnima | heh ;) |
| 02:18.07 | NonToxic | PaxAnima: k |
| 02:18.16 | PaxAnima | on the ipaq cluster |
| 02:18.23 | NonToxic | k |
| 02:18.56 | NonToxic | now we have to get the eon people to recompile using libsdl-jw so we can have bust'em and stuff on GPE :) |
| 02:19.13 | PaxAnima | heh.. |
| 02:19.31 | PaxAnima | so the eon games are just sdl? |
| 02:25.33 | PaxAnima | time to get some sleep... |
| 02:25.37 | PaxAnima | good night |
| 02:38.17 | nelson | NonToxic: what's sfcave? |
| 02:55.36 | NonToxic | nelson: what do you mean, 'what's sfcave?!' |
| 03:22.00 | *** join/#gpe flypiper (~deb-user@12-222-141-210.client.insightBB.com) |
| 03:22.00 | *** join/#gpe pigeon (~pigeon@CPE-203-45-112-207.nsw.bigpond.net.au) [NETSPLIT VICTIM] |
| 03:22.00 | *** join/#gpe jg (~jg@zap.crl.dec.com) [NETSPLIT VICTIM] |
| 03:22.00 | *** join/#gpe thr33 (olly@oghm2.acad.cai.cam.ac.uk) [NETSPLIT VICTIM] |
| 03:22.00 | *** join/#gpe cmarqu (colin@buug.mind.de) |
| 03:37.14 | nelson | NonToxic: by asking 'what's sfcave?' I mean that I don't know what sfcave is, and also that I'd like to know. |
| 03:40.02 | NonToxic | heh |
| 03:40.06 | NonToxic | hold |
| 03:40.17 | NonToxic | ibot sfcave |
| 03:40.17 | | well, sfcave is http://sfcave.com or http://sfcave.com/zaurus.php or http://andyq.myby.co.uk/ |
| 03:41.05 | NonToxic | crappy java game here |
| 03:41.06 | NonToxic | http://www.liquidcode.org/worm.html |
| 03:55.21 | *** join/#gpe prpplague (~joebob@22-30.lctv-b4.cablelynx.com) |
| 07:23.35 | *** join/#gpe zZzToxic (~joshua@got.owned.by-the.us) [NETSPLIT VICTIM] |
| 07:23.35 | *** join/#gpe chrisime (~chrisime@pD9E8CA79.dip0.t-ipconnect.de) [NETSPLIT VICTIM] |
| 07:23.35 | *** join/#gpe BBrox_ZZzZzzz (~ulmer@bbrox.org) [NETSPLIT VICTIM] |
| 07:28.06 | *** join/#gpe __moray (moray@dsl-217-155-67-241.zen.co.uk) [NETSPLIT VICTIM] |
| 08:16.53 | *** join/#gpe cian (~cian@reed.mle.ie) |
| 08:54.23 | *** join/#gpe mallum (~mallum@pc-80-193-218-21-hw.blueyonder.co.uk) |
| 10:13.43 | *** join/#gpe HET2 (~diman@chello080110105196.509.15.vie.surfer.at) |
| 10:24.57 | *** join/#gpe PaxAnima (~petter@v058b.studby.ntnu.no) |
| 10:31.43 | PaxAnima | morning |
| 11:05.44 | *** join/#gpe matire (~matire@h167n1c1o299.bredband.skanova.com) |
| 11:06.27 | *** part/#gpe matire (~matire@h167n1c1o299.bredband.skanova.com) |
| 11:35.03 | *** join/#gpe chrisime (~chrisime@pD9E8CA79.dip0.t-ipconnect.de) |
| 12:31.49 | *** join/#gpe paul ([3EXfdBWtX@212.17.60.230) |
| 13:31.43 | nelson | PaxAnima: god mor |
| 13:33.02 | PaxAnima | nelson: god morgen |
| 13:34.38 | mallum | hey |
| 13:34.59 | PaxAnima | hey, mallum |
| 13:38.54 | mallum | trying to get gprs enabled on phone .... |
| 13:39.56 | PaxAnima | mallum: what mobile phone do you have? |
| 13:40.10 | *** join/#gpe cian (~cian@reed.mle.ie) |
| 13:46.52 | mallum | PaxAnima: p800 |
| 13:47.19 | PaxAnima | ah... mr Gadget buys again I see... ;) |
| 13:47.33 | mallum | hehe |
| 13:52.25 | mallum | gprs sucks bigtime |
| 13:53.13 | PaxAnima | oh? |
| 14:01.02 | mallum | dont seem to work :\ |
| 14:03.32 | BBrox | mallum: which phone ? |
| 14:03.39 | mallum | BBrox: p800 |
| 14:04.40 | BBrox | So what's the problem ? |
| 14:04.51 | cian | mallum: supposed to be a nice phone alright |
| 14:05.16 | PaxAnima | mallum: have you enabled gprs for your mobile phone subscription? |
| 14:05.20 | mallum | BBrox: fucks knows :\, seemed to be able to connect but is really slow |
| 14:05.35 | mallum | PaxAnima: yeah I just got the magic text message to set it up |
| 14:05.39 | PaxAnima | ah |
| 14:05.40 | cian | mallum: and horribly expensive I think |
| 14:05.57 | cian | mallum: might be worth using phone just as normal modem |
| 14:06.51 | mallum | cian: yeah thats the plan |
| 14:06.54 | BBrox | mallum: when you say 'slow', what do you get as throughput ? |
| 14:07.03 | mallum | BBrox: 0.2k in 1 minute |
| 14:07.16 | BBrox | mallum: ah... yeah, kinda slow. |
| 14:08.31 | mallum | I'll try again later |
| 14:09.26 | BBrox | Did you try tcpdumping the IP stuff to see what's happening ? |
| 14:13.13 | mallum | BBrox: this is just with the browser on the phone |
| 14:13.21 | mallum | BBrox: I will try later over bluetooth |
| 14:15.53 | BBrox | mallum: ah :-) Never used GPRS otherwise than as a modem :-) |
| 16:09.26 | *** join/#gpe herzi (~herzi@arachni.kiwi.uni-hamburg.de) |
| 16:23.25 | *** join/#gpe pb_ (~pb@pc2-cmbg4-3-cust239.cmbg.cable.ntl.com) |
| 16:40.22 | *** join/#gpe dc_ (~dc@modem-39-54-60-62.vip.uk.com) |
| 16:48.26 | dc_ | afternoon |
| 16:49.05 | PaxAnima | hi, dc_ |
| 17:07.28 | *** join/#gpe jg_ (~jg@dhcp-174.crl.dec.com) |
| 17:22.11 | *** join/#gpe mallum (~mallum@pc-80-193-218-21-hw.blueyonder.co.uk) |
| 17:31.42 | pb_ | dc_: afternoon |
| 17:36.59 | *** join/#gpe walters (walters@verbum.org) |
| 17:43.43 | dc_ | afternoon pb_, mallum and walters |
| 17:49.11 | *** join/#gpe florian (~fuchs@pD9E5FD63.dip.t-dialin.net) |
| 17:49.19 | florian | hi all |
| 17:49.43 | pb_ | hi florian |
| 18:28.16 | *** join/#gpe dc__ (~dc@modem-35-43-60-62.vip.uk.com) |
| 18:55.19 | *** join/#gpe ade|tiny (~adrian@unimatrix.ex.ac.uk) |
| 18:56.53 | ade|tiny | ibot: date |
| 18:56.54 | | Thu May 1 18:56:54 2003 |
| 18:57.03 | ade|tiny | ibot: busybox date |
| 18:57.04 | | no idea, ade|tiny |
| 18:57.08 | ade|tiny | ibot: busybox |
| 18:57.09 | | hmm... busybox is at http://www.busybox.net It compiles with no trouble on a skiff or a single binary that replaces some 70 linux commands. things from ls to modprobe to vi to ... |
| 18:58.17 | pb_ | ibot, you forgot to say that it replaces them with versions that work much worse. |
| 18:58.17 | | You are moron #3, pb_ |
| 18:58.21 | pb_ | doh |
| 19:10.12 | paul | hi pb_ |
| 19:10.23 | dc__ | hey mr paul |
| 19:10.30 | paul | hey dc__! |
| 19:19.55 | pb_ | hi paul |
| 19:20.00 | paul | hi pb |
| 19:20.10 | paul | how be things? |
| 19:20.49 | pb_ | yeah, not bad. been kind of busy the last couple of days. |
| 19:20.53 | paul | pb_: Xipaq memory usage is so so much better these days. is that the change to libc (the 'use mmap' point?) |
| 19:21.25 | pb_ | paul: not sure. I don't think the server itself has changed, though check with jg. |
| 19:23.36 | paul | had it on a few days now, and its 5.9MB. about 2.5MB in data, 2 in libs, there's about 0.4MB-(sizeof(framebuffer)) unaccounted for which must be private mappings. |
| 19:24.00 | pb_ | right. can check /proc/pid/maps to get the details. |
| 19:25.06 | mallum | paul: you can play with how the xserver sets its nmap threshhold thingy, by setting various env vars |
| 19:25.52 | paul | mallum: ah really? |
| 19:26.00 | paul | mallum: that new? |
| 19:26.19 | paul | (and thats the var do you know, to save me looking through output of strings) |
| 19:26.36 | mallum | paul: no its a feature of glibc |
| 19:27.09 | mallum | paul: see http://www.linuxjournal.com/article.php?sid=6390 |
| 19:27.25 | paul | mallum: ah.. ok. |
| 19:27.58 | mallum | paul: see http://www.linuxjournal.com/modules.php?op=modload&name=NS-lj-issues/issue109&file=6390t1 also |
| 19:29.56 | paul | mallum: ah yes... that's in the glibc info page too. |
| 19:30.16 | jg | paul, pb_: the server hasn't changed significantly. The change to libc may be the help... |
| 19:30.21 | paul | mallum: might be worth experimenting to see how performance is affected as that setting is lowered. |
| 19:30.29 | paul | jg: yes, i thought so. |
| 19:31.18 | jg | paul: the setting *MUST* be bigger than the most complicated region structures get, or performance (and as a result, power consumption) will plummet.... |
| 19:32.22 | paul | jg: is that because of slowness of allocation? |
| 19:32.27 | jg | yup. |
| 19:32.56 | jg | regions are manipulated, created, and destroyed with gay abandon... So having to do system calls to do each one becomes a performance issue. |
| 19:33.01 | paul | jg: so there should be a reasonably well-defined point where performance drops. |
| 19:33.17 | jg | yes, likely related to the physical size of the screen. |
| 19:45.34 | paul | jg: so with enough experimentation you could find the optimum screen size scalar for the mmap() point... |
| 19:46.46 | jg | paul: yes, but some computation makes me think that 16K is probably OK; below that, I get nervous.... |
| 19:46.59 | jg | a region structure is about 12 bytes in size.... |
| 19:47.23 | jg | So that allows a complexity of about 1500 rectangles. |
| 19:47.26 | paul | jg: and those are the biggest structures? |
| 19:48.09 | jg | no, pixmaps are the biggest structures, but they don't get created/destroyed at the rate region structures do, which is very frequent when you do alot of window manipulation. |
| 19:48.15 | paul | ah. |
| 19:48.48 | jg | you *want* as many pixmaps to get allocated this way as possible, as then when they are destroyed, you are able to give the memory back.... |
| 19:49.13 | jg | paul: 8k might concievably be ok, but you'd have to try it out. |
| 19:49.51 | paul | even if a structure and the data it points to were to approach 16K, the struct and individual bits of data would be allocated one by one, and hence be lower than 16k to a significant degree. (assuming several pieces of data) |
| 19:50.24 | paul | jg: yes, thats what i thinking of when you first explained the pixmap/client request problem and i read the glibc info pages on this. |
| 19:51.15 | jg | no, the region code allocates and frees structures the size of the array of rectangles required to represent the region. |
| 19:51.56 | pb_ | paul: the other problem with making the threshold very small is that wastage becomes significant. All mmapped allocations are rounded up to a multiple of 4k. |
| 19:52.30 | paul | jg: conceivably X could class client-supplied/influenced data 2 ways 'frequently' and 'infrequently' allocated (the former being performance sensitive in allocation) and X could manually mmap() memory for the latter class. (if object were considered big enough) |
| 19:52.52 | paul | pb_: presumably the wastage is in whatever data structures glibc must use to manage the heap? |
| 19:53.03 | paul | pb_: oh... mapped allocations, ok. |
| 19:53.14 | jg | paul: the basic improvement we need to make some day is to separate pixmap allocation from other heap allocation of the server. This needs to be done anyway.... |
| 19:54.06 | jg | Then this wouldn't be an issue, and when clients exit and lots of pixmaps go away, we could pretty likely recover the address space used quite nicely. |
| 19:54.08 | paul | jg: and instead allocate them via ..... (mmap?) ? |
| 19:54.17 | paul | ah ok. |
| 19:54.43 | jg | paul: or whatever: the point being that it would be good on some displays to be putting the pixmaps out in the offscreen memory of the display anyway. |
| 19:55.05 | paul | jg: the actual video memory? |
| 19:55.10 | jg | yup. |
| 19:55.18 | jg | So this separation should be done anyway. |
| 19:55.36 | paul | jg: damn, do graphics cards have /that/ much memory these days :) (i only have 8MB G200) |
| 19:55.48 | jg | some have 128 meg.... |
| 19:56.22 | paul | jg: does XAA not do this to some extent already (i know nothing about XAA, but i thought it tried to use extra video ram in some way) |
| 19:56.48 | jg | it has code for shoving the pixmaps back and forth; but not for the initial allocation. |
| 19:56.53 | jg | (or so I understand). |
| 19:57.29 | paul | ah... so they're shadowed to video ram. and in future X migth try avoid the shadowing in system ram if it can. sounds good :) |
| 19:57.36 | jg | yup. |
| 19:58.00 | paul | that would be nice to have. |
| 19:59.36 | paul | anyway, with the glibc change it /seems/ Xipaq is using slightly less RAM.. for moment i'm presuming this is because some stuff is now being allocated via mmap which before would have gone via heap. i'll try do some heavy internet browsing tonight when i get home with dillo and see what happens. |
| 20:00.01 | pb_ | ok, cool. |
| 20:00.03 | paul | being allocated and freed. |
| 20:00.39 | paul | now.... tea and back to script writing :) |
| 20:00.43 | dc__ | hehe |
| 20:23.26 | paul | oh.. about machines with huge amounts of video ram, eg 128MB.. well, i have a machine somewhere with about 700MB of video ram. |
| 20:24.08 | paul | (an SGI O2 which has uses main memory for video memory :) ) |
| 20:25.04 | mallum | the xserver on O2's rules |
| 20:26.13 | paul | mallum: yeah? |
| 20:26.27 | paul | mallum: that wolf picture demo is awesome :) |
| 20:26.39 | paul | the one where you can make it ripple like water. |
| 20:26.55 | paul | or grab it with the mouse and stretch it like its rubber. |
| 20:27.13 | mallum | yeah, and the metalic case thingy |
| 20:27.21 | paul | metalic case thingy? |
| 20:27.51 | mallum | yeah there is a demo with a reflective 02 case you can rotate around |
| 20:28.27 | paul | mallum: hmm.. havnt seen that one. |
| 20:29.26 | paul | mallum: linux support for O2 is apparently coming along now, there's even the beginnings of support for the O2's ICE chip, the compression engine. |
| 20:32.29 | mallum | paul: ah cool |
| 20:32.48 | paul | amazing how cheap O2's are on ebay these days.. |
| 20:33.41 | *** join/#gpe BBrox (~ulmer@bbrox.org) |
| 20:42.18 | *** join/#gpe flypiper (~deb-user@12-222-141-210.client.insightBB.com) |
| 22:34.40 | *** join/#gpe dc__ (~dc@modem-35-43-60-62.vip.uk.com) |
| 22:41.13 | *** join/#gpe walters (walters@verbum.org) |
| 23:20.14 | *** join/#gpe ade|chloe (~adrian@unimatrix.ex.ac.uk) |
| 23:54.19 | *** join/#gpe jg_ (~jg@h005018015b26.ne.client2.attbi.com) |