00:34.54 | *** join/#picogui scanlime (micah@aden2-42-dhcp.resnet.Colorado.EDU) |
00:39.00 | captain_proton | and for some reason, when i start it up it thinks i've got the control key pressed |
00:39.10 | scan[pterm] | hmm |
00:39.32 | captain_proton | eep |
00:39.44 | scan[pterm] | oh |
00:39.49 | scan[pterm] | it probably thinks alt is pressed |
00:39.55 | captain_proton | alt? |
00:39.58 | scan[pterm] | yes |
00:40.00 | captain_proton | its acting like control |
00:40.08 | captain_proton | ie: press d, it logs out |
00:40.17 | scan[pterm] | hmm |
00:40.24 | scan[pterm] | if you press control and release it, does that fix it/ |
00:40.33 | scan[pterm] | s/\//?/ |
00:40.34 | captain_proton | yeah |
00:40.49 | scan[pterm] | I've seen something similar happen if I have the terminal focused and change virtual desktops |
00:40.56 | scan[pterm] | bug in the x11input driver |
00:41.07 | captain_proton | another problem: i'm not getting the titlechange event |
00:41.11 | scan[pterm] | oh? |
00:41.30 | scan[pterm] | have you tried the regular pterm too, to make sure your copy of the server is sending the event? |
00:42.16 | captain_proton | hm, apparently its not |
00:42.29 | scan[pterm] | cvs up pgserver and recompile it? :) |
00:42.38 | captain_proton | done that |
00:42.55 | scan[pterm] | and for pterm I assume? |
00:43.08 | scan[pterm] | pterm was modified to set TERM=xterm |
00:43.14 | scan[pterm] | besides the actual event handler |
00:43.28 | captain_proton | yeah, have that |
00:48.41 | captain_proton | hrm |
00:48.45 | captain_proton | new build made no difference |
00:48.51 | captain_proton | you committed that code, right? :P |
00:48.53 | scan[pterm] | hmmmmmm |
00:49.20 | scan[pterm] | yeah, only uncommitted code I have is a little debugging for the x11 mouse reporting |
00:49.37 | scan[pterm] | and without the new constants.h, pterm wouldn't have compiled |
00:51.41 | scan[pterm] | are you sure your shell is sending the title change? |
00:51.53 | scan[pterm] | you could install vttest and use it to send the title change |
00:52.44 | captain_proton | should be |
00:53.01 | scan[pterm] | try vttest |
00:53.53 | captain_proton | ugh |
00:53.57 | captain_proton | galeon seggie |
00:54.34 | captain_proton | argh |
00:54.35 | captain_proton | did again |
00:56.56 | captain_proton | hm, that worked |
00:58.29 | lalo[out] | captain_proton: you should probably run the user's login shell |
00:58.56 | captain_proton | lalo[out]: yeah yeah... |
00:59.01 | lalo[out] | not very terribly hard to do in python |
00:59.31 | scanline | methinks the picogui launcher I'll be writing for 1.0 should be in python |
01:00.08 | scanline | it will be dramatically easier to write all the launcher and configuration infrastructure in python, and if the VR3 can handle it, anything people would need a launcher on should be able to :) |
01:00.39 | captain_proton | scanline: is there a way for the tabpage to truncate its text instead of making the window grow? :P |
01:00.51 | scanline | captain_proton: not yet, you'd have to do it yourself |
01:00.52 | lalo[out] | I considered writing a launcher in python |
01:01.14 | lalo[out] | problem is it consumes a lot of ram, specially in the vr3 |
01:01.26 | scanline | ah |
01:01.37 | captain_proton | it would be nice if there was some way to make python share all the core-type bits |
01:01.42 | lalo[out] | yup |
01:01.46 | lalo[out] | I mean |
01:01.53 | lalo[out] | there is :-) |
01:02.13 | scanline | how about if it's launching another python app, instead of calling python it imports the file and runs it in a thread? |
01:02.18 | lalo[out] | but it's clumsy |
01:02.27 | lalo[out] | scanline: that's exactly what I was thinking |
01:02.28 | scanline | hmm, that's probably get slow with python's not-really-threadedness |
01:02.47 | scanline | worth a try though |
01:02.59 | lalo[out] | no, python threads aren't too slow |
01:03.05 | lalo[out] | zope works, after all :-) |
01:03.15 | scanline | lalo[out]: I was just thinking about the big interpreter lock :) |
01:03.18 | lalo[out] | but the f*ing snow environment has no threads |
01:03.37 | scanline | ah |
01:03.40 | lalo[out] | the interpreter lock is no problem... it is "thread-safe" in *this* aspect |
01:03.49 | lalo[out] | I mean |
01:03.58 | scanline | well, I know it's thread-safe. but the huge lock I thought would make things slow |
01:03.59 | lalo[out] | scratch that, my brain dozed |
01:04.24 | scanline | well, I could write us a good launcher in C |
01:04.27 | lalo[out] | I was thinking of cli_python's event loop |
01:05.03 | lalo[out] | I agree a python launcher would be cool |
01:05.05 | scanline | but it would be really easy to do a nice launcher and configuration app in python |
01:05.24 | lalo[out] | and the only place where it's a problem is the f*ing braindead snow |
01:05.27 | scanline | hmm... |
01:05.29 | scanline | another decision: |
01:05.42 | scanline | should app-server be a 1.0 goal, or should the launcher for 1.0 use something like res_c? |
01:05.50 | lalo[out] | I would anyway if I went uclibc |
01:06.23 | lalo[out] | not sure... I want an app server... but I also want to get rid of this and start working on 2.0 :-P |
01:06.35 | scanline | yeah |
01:07.06 | scanline | at this point I would lean toward the option that gives us less support hassle while we're trying to code 2.0 :) |
01:07.29 | scanline | and that would be a python module for managing app packaging, and a launcher written in python that uses it |
01:07.51 | captain_proton | scanline: what is the property in Application for the title? |
01:08.02 | scanline | PG_WP_TEXT |
01:08.13 | scanline | er, 'text' in python |
01:09.59 | captain_proton | mind if i quickly flood? |
01:10.04 | scanline | I don't mind |
01:10.07 | captain_proton | Traceback (most recent call last): |
01:10.07 | captain_proton | <PROTECTED> |
01:10.07 | captain_proton | <PROTECTED> |
01:10.07 | captain_proton | <PROTECTED> |
01:10.07 | captain_proton | <PROTECTED> |
01:10.08 | captain_proton | <PROTECTED> |
01:10.10 | captain_proton | <PROTECTED> |
01:10.12 | captain_proton | <PROTECTED> |
01:10.14 | captain_proton | <PROTECTED> |
01:10.16 | captain_proton | TypeError: tabClicked() takes exactly 2 arguments (3 given) |
01:10.59 | scanline | hrm.. |
01:11.07 | lalo[out] | bug |
01:11.33 | lalo[out] | that means your code raised a TypeError, and the dispatcher innocently thought you were requesting for a third argument |
01:11.53 | lalo[out] | perhaps we bork less code if we remove this test and the DeprecationWarning? :-P |
01:12.04 | scanline | I'd leave it in |
01:12.45 | scanline | though it would be nice if it could tell when an exception was about the arguments and when it's not |
01:15.29 | captain_proton | yay for updates :) |
01:15.36 | captain_proton | epterm is starting to almost be usable |
01:16.00 | njs | there's no way to use python introspection to ask how many arguments a callable takes? |
01:16.40 | lalo[out] | njs: kinda |
01:17.05 | njs | (of course, in python, a callable can take either 2 or 3 arguments, and I don't know what you plan to do then :-)) |
01:17.27 | lalo[out] | but it breaks if the handler is a method of a class, or an object pretending to be a function |
01:17.32 | lalo[out] | so it's not worth it |
01:25.20 | *** join/#picogui kiatoa (~matt@rich-dsl3-66.greenmountainaccess.net) |
01:28.50 | kiatoa | knock, knock, anyone home? Scanline? |
01:33.00 | lalo[out] | 'nite |
01:33.19 | kiatoa | bye |
01:53.16 | scanline | kiatoa: hi |
01:53.27 | *** join/#picogui rev|desk (Malkin@umd97-21.d.umn.edu) |
01:53.36 | rev|desk | ho-la |
01:53.43 | scanline | hi rev|desk |
01:54.37 | rev|desk | and i must run picogui on it at some point |
01:54.45 | scanline | :) |
01:55.33 | rev|desk | i was playing with the poor helio+pgui again |
01:55.40 | scanline | ha |
01:55.54 | rev|desk | i had some grand plans based on a simple squeak app |
01:56.00 | rev|desk | but gave up |
01:56.08 | rev|desk | i wish someone wanted it, even for like $10-30 |
01:56.17 | rev|desk | just so that it was being *used* instead of just sitting in a cubbord |
01:56.37 | scanline | I have a couple helios doing nothing as well... |
01:56.42 | rev|desk | a shame |
01:56.50 | scanline | might try to update the picogui and linux on them once I have some usable software |
01:56.53 | rev|desk | i could make a beowulf cluter of them |
01:56.54 | rev|desk | yeah |
01:57.12 | rev|desk | i've been using it under VT-OS some lately as a watch, and calendar |
01:57.22 | rev|desk | just because the newton can be huge in the pocket sometimes |
01:57.27 | rev|desk | but with this zaurus |
01:57.30 | rev|desk | it's small enough |
01:57.54 | scanline | yep |
01:57.55 | scanline | bbiab |
01:58.00 | rev|desk | like, i usually have my newton with me during the day at work and in class, but if i'm going out, or something, i hate to bring the big newton just for a time piece |
01:59.07 | *** join/#picogui gonkulator (~brandon@aden2-235-dhcp.resnet.colorado.edu) |
01:59.24 | gonkulator | hi |
02:00.07 | rev|desk | hey gonkie |
02:00.21 | captain_proton | scanline: food\ |
02:07.12 | *** join/#picogui file (file@mctn1-7519.nb.aliant.net) |
02:26.26 | *** part/#picogui rev|desk (Malkin@umd97-21.d.umn.edu) |
02:40.58 | *** join/#picogui gonkulator (~brandon@aden2-244-dhcp.resnet.colorado.edu) |
02:41.53 | *** join/#picogui decaff (khedspet@1Cust81.tnt20.tpa2.da.uu.net) |
02:42.27 | *** join/#picogui file (file@mctn1-7147.nb.aliant.net) |
02:58.33 | *** join/#picogui prpplague (~joebob@21-190.lctv-b4.cablelynx.com) |
02:59.38 | *** join/#picogui decaff (khedspet@1Cust81.tnt20.tpa2.da.uu.net) |
03:26.05 | captain_proton | scanline: care to do some stuff? :P |
03:26.19 | scanline | eh? |
03:26.37 | captain_proton | 1) tabpage |
03:27.01 | captain_proton | 2) modified hotkeys |
03:30.04 | scanline | ehmyeah... |
03:30.11 | scanline | I'm still doing the VT102 emulation :P |
03:30.40 | njs | scanline: how does picogui handle "scheduling"? |
03:31.01 | scanline | not very well :) |
03:31.15 | njs | scanline: meaning, you're using an event loop, so each time around you have to decide what to do next, and the decision matters... |
03:31.29 | scanline | right |
03:31.52 | scanline | since speed there hasn't been a problem, and that code's going to get rewritten anyway, it just goes in order of file descriptor |
03:32.01 | njs | round-robin? |
03:32.05 | scanline | yes |
03:32.20 | njs | and a redraw in between each round, or? |
03:32.30 | scanline | no, a redraw when one is necessary |
03:32.54 | njs | well, "redraw" = "repair any damage (doing nothing if there is no damage)" in my head :-) |
03:32.56 | captain_proton | scanline: thats not based on birds |
03:32.57 | scanline | picogui defines that as after something has changed, and before user interaction is required |
03:33.02 | captain_proton | its based on robin, king of the britains |
03:33.11 | scanline | captain_proton: ah |
03:33.15 | scanline | then that's ok |
03:33.19 | njs | captain_proton: is it really? huh. I always wondered about that. |
03:33.36 | njs | scanline: well, you need some word... "in order of file descriptor" could mean handle all outstanding requests on the first before going on to the second :-) |
03:33.45 | scanline | njs: ah. yeah |
03:34.08 | scanline | njs: but that code has been given hardly any thought.. I know it needs smarter scheduling, but there have been more important things to do |
03:34.32 | captain_proton | like...tabpage bugfixes! |
03:34.43 | scanline | ...or the vt102 emulation... |
03:34.56 | captain_proton | silly silly emacs user |
03:37.23 | njs | I wonder if there's ever any performance benefit to handling multiple requests from a single client at once. |
03:37.53 | njs | In particular, I wonder if keeping the cache warm by doing that is a measurable win. |
03:38.57 | scanline | would be worth profiling |
03:39.22 | scanline | picogui spends very nearly all its CPU time in the video driver though |
03:39.30 | njs | seems possible, especially in something like v2's architecture, where resolving resource ids -> objects implementing them via some per-client table would be a common operation, and possibly a bit expensive. |
03:40.12 | njs | (so keeping the table in cache would be good) |
03:40.23 | njs | yeah, of course |
03:41.12 | njs | I wonder how much of that is necessary -- i.e., if you had a proper acceleration architecture and were just shoving stuff into the card's render queue. |
03:42.27 | scanline | would be nice if we had any acceleration architecture :) |
03:42.38 | njs | well, yes :-) |
03:43.04 | njs | but architecting on the assumption that it will exist eventually is reasonable :-) |
03:44.54 | njs | (Fresco currently spends some huge percentage of its time just in compositing and blitting to the framebuffer.) |
03:46.18 | njs | (well, that's true when using libart to render; I haven't profiled GLX) |
04:42.08 | scanline | <scanline> captain_proton: all the bugs you're complaining about should be scheduled for the 0.46 release |
04:42.11 | scanline | <scanline> so hopefully we can do a release of picogui that includes many bugfixes, plus epterm and nifty |
04:42.17 | scanline | there, correct channel :) |
04:42.23 | Xentac | hehehe |
04:43.36 | captain_proton | =) |
04:45.42 | scanline | hmm.. 0.45 needs a release name like "Strapping the first boot", since it will be possible to work on picogui from within picogui |
04:49.28 | njs | it handled vi before this round of hacking even started. |
04:49.53 | Xentac | hehehe |
04:49.57 | Xentac | figured as much |
04:51.37 | scanline | well, some types of vi |
04:51.45 | scanline | busybox vi worked, vim had problems with scrolling |
04:51.50 | scanline | now it should all work |
04:51.59 | Xentac | how much vt102 emulation is actually implemented? |
04:52.37 | scanline | very nearly all of it |
04:52.50 | scanline | but it was pretty much all implemented before, just really buggy |
04:53.15 | Xentac | hehehe |
04:54.32 | scanline | launcher :) |
04:54.38 | Xentac | hehehe |
04:54.40 | scanline | or fix bugs, they're small :) |
04:55.12 | njs | "Take two, they're small." |
04:55.57 | Xentac | problem with bugs is that I don't really know where to start ;o) |
04:56.10 | scanline | Xentac: there are some simple things to do also... check the "Ask not what PicoGUI can do for you..." news item |
04:56.28 | scanline | one that pops into my mind is to move bugs from the old database to the new one :) |
04:56.41 | Xentac | hehehe, yeah... but that's so boring :P |
04:57.12 | scanline | yeah... he could also copy over the old wiki entries |
04:57.15 | scanline | maybe file :) |
04:57.21 | Xentac | that's who I was going to suggest |
04:58.41 | Xentac | scanline: looks like page up and page down don't work in vim... but scrolling looks fine |
04:58.47 | scanline | hmm |
04:58.51 | scanline | page up and page down work for me... |
04:59.07 | Xentac | in vim? |
04:59.07 | scanline | press ctrl and alt once each, you may be getting bit by a bug in the x11input driver where modifiers get stuck down |
04:59.24 | Xentac | nope, still does it |
04:59.29 | scanline | oh |
04:59.39 | scanline | hrm, guess I broke it when trying to implement the prefix switch |
04:59.58 | scanline | gives me a "5~" or "6~" on the screen |
05:00.49 | captain_proton | 'stuff!' |
05:01.06 | scanline | sounds like the jetstream repository :) |
05:01.11 | captain_proton | uhhuh |
05:01.22 | captain_proton | scanline: care to make sure epterm does all its goodness on your machine? |
05:01.28 | scanline | k |
05:02.20 | scanline | you could have used a message like "Improved dynamic title support" |
05:02.25 | captain_proton | hehe |
05:02.27 | captain_proton | but you see |
05:02.34 | captain_proton | for that, i would have had to remember what i changed since my last commit |
05:02.38 | captain_proton | or...done a diff |
05:02.42 | scanline | that's what "cvs diff" is for |
05:02.44 | scanline | heh |
05:02.52 | captain_proton | pah! |
05:02.59 | scanline | but yeah, CVS is slow... |
05:03.39 | captain_proton | yupyup |
05:03.44 | scanline | hmm, why did I just run 'make' in the epterm directory... |
05:04.52 | Xentac | what are you going to use instead of auto*? |
05:04.54 | scanline | yay for tabbiness |
05:04.58 | scanline | Xentac: probably SCons |
05:05.05 | scanline | plus a lot of new code for configuration |
05:05.06 | captain_proton | or |
05:05.11 | captain_proton | a big-assed bash script! |
05:05.18 | scanline | most of that stuff will migrate pretty easily to picogui 2.0 when that time comes |
05:05.22 | Xentac | hehehe... most of arch's scripts are written in bash |
05:05.57 | scanline | hmm, so function keys only work for the first two tabs? |
05:06.35 | captain_proton | scanline: for now |
05:06.52 | Xentac | scanline: what exactly do you want done with the launcher? (I assume it's pgl we're talking about) |
05:07.28 | scanline | ah |
05:07.36 | scanline | I just committed a really quick fix anyway |
05:08.01 | scanline | Xentac: well, I think it makes more sense at this point to scrap pgl and write one in python |
05:08.10 | Xentac | ah, I see |
05:08.50 | Xentac | and I'm still confused as to what you want it to do... like... it's a menu bar... |
05:09.03 | Xentac | that leaves a lot open to interpretation ;o) |
05:09.08 | scanline | not necessarily |
05:09.32 | scanline | I was planning on the launcher being based on an embeddable file manager |
05:09.34 | njs | playing with byte swapping code, I just timed my trivial macro versions compared to the inline asm versions in <byteswap.h>. |
05:09.49 | scanline | but without app-server, might as well just make it a set of python modules |
05:10.08 | Xentac | embeddable file manager, eh? that doesn't exist yet, does it? |
05:10.12 | Xentac | app-server...? |
05:10.18 | njs | And for shorts, my macro is 4.2 times slower than the asm. And for longs, my macro is 2.4 times _faster_ than the asm. |
05:10.19 | scanline | you'd have a module that would act like a file manager, and depending on the configuration it could appear in a menu-thingy, root menu, on the desktop... |
05:10.32 | scanline | njs: makes sense |
05:10.44 | scanline | :) |
05:10.45 | njs | scanline: how so? |
05:11.13 | Xentac | scanline: or all three...? |
05:11.25 | scanline | Xentac: yeah |
05:11.33 | njs | scanline: ah :-) |
05:12.03 | scanline | Xentac: the python module solution is easy and maintainable, but it locks anything that needs to use the file manager into python... |
05:12.27 | scanline | Xentac: but I don't think any of us want to put in the work to make app-server work when this is all eventually getting scrapped for 2.0 anyway |
05:12.30 | scanline | :) |
05:12.51 | captain_proton | explain this mystical app-server |
05:13.12 | scanline | It was going to be responsible for tasks that are common between clients, but couldn't be tied to the server machine |
05:13.21 | scanline | so you'd have one app-server running on each machine that had picogui clients active |
05:13.42 | scanline | it would be responsible for loading and reading application packages, mapping mime types to app packages, and running embedded apps |
05:13.47 | Xentac | silly sickness |
05:13.56 | scanline | g'night Xentac |
05:14.02 | Xentac | night |
05:18.38 | njs | Nope, looking at the asm sheds no light on this mystery. |
05:20.22 | captain_proton | njs: ghosts |
05:22.53 | scanline | ~sci-fi du jour |
05:22.57 | | Explosion imminent: inverse flux tachyon emission generator in engineering, ensign spock |
05:33.40 | njs | Hmm. There are times when it's advantageous to have a gcc hacker sitting on the floor in your living room. |
05:34.20 | *** join/#picogui darth_iBook (~wallacesh@64.denver-23-24rs.co.dial-access.att.net) |
05:35.05 | scanline | njs: :) |
05:35.11 | njs | the moral of the story is: bswap_32 > hand-hacked macro _iff_ you compile with -march=i486 or better. |
05:35.20 | scanline | ah |
05:36.03 | njs | because 386-compatible optimized version manages to cause a partial register stall (!). |
05:36.11 | scanline | haha |
06:01.59 | *** join/#picogui zak (noname@aden2-135-dhcp.resnet.Colorado.EDU) |
06:02.06 | zak | scanline: knock, knock |
06:02.41 | captain_proton | zak: whos there? |
06:03.11 | zak | captain_proton: me!!! |
06:03.24 | zak | me!!! who wants a soda! |
06:03.43 | captain_proton | zak: hmm |
06:03.56 | captain_proton | gonkulator left 7/8 of a cel-ray in scanline's room, i'm sure you could have that |
06:04.16 | zak | i knock knocked him in two channels and msg'd him....he is not paying attention |
06:04.17 | captain_proton | scanline: make gonkulator get his ibook, if he's off to bed he should take it with him |
06:08.28 | captain_proton | hmm |
06:08.29 | captain_proton | scanline |
06:08.39 | captain_proton | does x11 have problems if the window is larger than the desktop? :) |
06:46.42 | njs | not in general, no. |
06:46.54 | njs | though you may have some trouble using your WM to manipulate it |
06:47.04 | scanline | xfree86 doesn't like creating a pixmap larger than the screen |
06:47.21 | scanline | and by default, picogui's x11 driver works using SHM pixmaps |
06:47.49 | scanline | I plan on eventually fixing the driver to use many smaller pixmaps though, because xfree86 is terribly slow at manipulating large pixmaps |
06:47.55 | njs | ah, windows and pixmaps may be different |
07:02.37 | *** join/#picogui captain_proton (~jupiter@aden2-23-dhcp.resnet.Colorado.EDU) [NETSPLIT VICTIM] |
07:20.46 | njs | Heh. |
07:21.24 | njs | And in further adventures in glibc-land, our hero discovers that when your arrays aren't four-byte aligned, copying them with a simple for loop is faster than memcpy, by a good 40%. |
07:22.00 | scanline | I suspect that's only true on x86 |
07:22.16 | njs | why is that? |
07:22.40 | scanline | most other CPUs will give you a bus error for accessing unaligned data, or use a very slow OS trap to correct it |
07:22.57 | njs | (what I said is probably a bit misleading, incidentally; the for loop is copying 4 bytes at a time, by casting to a long[]) |
07:23.38 | scanline | oh.. so the beginning of the array is aligned on a 4-byte boundary? |
07:24.09 | njs | yes, I was talking about the alignment of the beginning of the array. |
07:24.20 | scanline | ok |
07:25.13 | njs | well, numbers: 8-byte aligned: for loop: 0.34s, memcpy: 0.27 seconds. 4-byte aligned: for loop: 0.34s, memcpy: 0.33s. Unaligned: for loop: 0.34s, memcpy: 0.5s. |
07:25.40 | scanline | strange |
07:25.48 | njs | (this is cpu time measured with clock(), copying 40,000,000 bytes on my PIII.) |
07:26.04 | scanline | I would have expected memcpy() to be the same in all three cases and the for loop to be slower when unaligned |
07:27.06 | njs | oh, hmm. the source array is 8-byte aligned in all cases; I wonder if that matters. |
08:02.33 | *** join/#picogui lurgyman (kuester@dialup-85-38.Colorado.EDU) |
09:08.25 | *** join/#picogui captain_pistachi (~jupiter@aden2-23-dhcp.resnet.Colorado.EDU) |
09:19.10 | *** join/#picogui gonkulator (~brandon@aden2-241-dhcp.resnet.colorado.edu) |
13:14.22 | *** join/#picogui gonkey{iBook} (~brandon@aden2-241-dhcp.resnet.colorado.edu) |
14:38.02 | *** join/#picogui file (file@mctn1-1468.nb.aliant.net) |
15:07.17 | file | Greetings Xentac |
15:07.43 | Xentac | morning file |
15:07.54 | file | elaborate |
15:08.26 | Xentac | I'm feeling sick... but I have to feel good for the tournament on saturday... |
15:08.40 | file | evilness :( |
15:09.03 | Xentac | quite |
15:09.10 | Xentac | I just started feeling sick tuesday |
15:11.42 | file | Xentac: :\ |
15:30.26 | *** join/#picogui prpplague (~JoeBob1@12.148.134.9) |
16:47.18 | *** join/#picogui Dentoid (dento_3000@h242n1fls35o814.telia.com) |
16:47.25 | Dentoid | Hi ho |
17:21.17 | gonkulator | file: how do you do that? |
17:21.58 | file | gonkulator: company had a deal on |
17:22.04 | file | first month costs a penny |
17:22.25 | gonkulator | oh |
17:22.31 | gonkulator | whats your toll-free number? |
17:22.40 | file | it hasn't been setup yet |
17:22.45 | file | takes 4-6 hours (I called and asked) |
17:22.50 | gonkulator | oh, ok |
17:23.16 | gonkulator | the part I like is the fact that it is a canadian penny, which is even more worthless than an american penny :) |
17:23.51 | file | it's USD :p |
17:24.36 | gonkulator | oh |
17:34.43 | kergoth | mmmmm ramen |
18:08.56 | lalo[out] | hi |
18:09.25 | file | hey lalo |
18:11.29 | *** join/#picogui nemo (nemo@61.95.53.30) |
18:22.10 | lalo[out] | scanline: you napping? |
18:22.27 | lalo[out] | WHAAAAAAA! NO PICOBOT! :~( |
18:23.03 | file | scanline has been away for 7 and a half hours. |
18:25.05 | file | lalo[out]: gonkulator is waking scanline up |
18:29.21 | lalo[out] | why? |
18:29.25 | lalo[out] | let the man sleep :-) |
18:41.41 | *** join/#picogui zak (~noname@aden2-35-dhcp.resnet.Colorado.EDU) |
18:54.11 | *** join/#picogui prae (~prae@sherpadown.net) |
19:11.03 | *** join/#picogui Dentoid (dento_3000@h242n1fls35o814.telia.com) |
19:14.46 | gonkulator | lalo[out]: scanline told me to get rid of PicoBot |
19:14.53 | gonkulator | lalo[out]: do you want him back? |
19:17.28 | lalo[out] | if scanline doesn't want him, he probably has a reason... I can argue with him later |
19:17.45 | scanline | I don't not want him, just didn't think he was necessary |
19:17.57 | file | scanline: what did you think of the scary news? |
19:18.16 | scanline | ibot does all that picobot does and more.. |
19:18.16 | | scanline: I wish you would RTFM. |
19:18.18 | scanline | file: scary news? |
19:18.26 | scanline | ibot: oh get stuffed |
19:18.27 | | scanline: I don't know, could you explain it? |
19:18.38 | file | scanline: me, Gentoo |
19:18.50 | file | installing it... |
19:19.12 | scanline | oh |
19:19.13 | scanline | scary |
19:20.12 | lalo[out] | oh sure, I keep forgetting ibot |
19:20.19 | lalo[out] | ibot: seen scanline |
19:20.20 | | scanline is currently on #picogui (1d 5h 33m 8s) #elinux (1d 5h 33m 8s) #openembedded (1d 5h 33m 8s). Has said a total of 208 messages. Is idling for 1m 7s |
19:20.24 | lalo[out] | yup. |
19:20.27 | lalo[out] | ibot: explode |
19:20.27 | | i heard explode is no |
19:20.35 | lalo[out] | hmm, perhaps not everything. |
19:20.41 | scanline | true |
19:22.35 | lalo[work] | let's see if the nick can influence my brain |
19:22.35 | scanline | have fun at [work] :) |
19:22.45 | scanline | hmm |
19:23.42 | lalo[work] | hahahahahahaha |
19:24.11 | scan[productive] | hmm, nope |
19:24.22 | scan[frustrated] | hey, that one works fine |
19:28.35 | file | mmm I'll have Gentoo tonight, nifty |
19:28.57 | lalo[work] | there isn't an ebuild for nifty yet |
19:29.07 | scanline | hey, the M3 meeting for Fresco is starting soon, if anyone's interested |
19:29.12 | file | gonkulator: will you be around to assist me? |
19:29.20 | gonkulator | M3? |
19:29.27 | scanline | milestone 3 |
19:29.30 | gonkulator | oh, ok |
19:30.13 | gonkulator | file: you should be able to if you can RTFM |
19:30.27 | file | gonkulator: :) |
19:31.21 | *** join/#picogui DevGirl (maisa@200.203.12.232) |
19:31.27 | DevGirl | hello, hello |
19:31.36 | scanline | hiya DevGirl |
19:31.43 | DevGirl | OH! NO PICOBOT!!!!! |
19:31.44 | file | DevGirl! |
19:31.53 | scanline | :( |
19:32.00 | DevGirl | hiya, scanline, file |
19:32.03 | DevGirl | :P |
19:32.12 | DevGirl | who will explode for mr? |
19:32.15 | DevGirl | me? |
19:32.17 | DevGirl | :( |
19:32.32 | scanline | ibot: explode is <action> explodes into a happy puff of smoke |
19:32.34 | | ACTION blows up is <action> explodes into a happy puff of smoke with bombs |
19:32.43 | scanline | ooh |
19:32.44 | lalo[work] | hahahahahaha |
19:32.45 | scanline | ibot: explode |
19:32.46 | | i guess explode is no |
19:32.49 | scanline | hmm |
19:32.52 | scanline | ibot: forget explode |
19:32.52 | | i forgot explode, scanline |
19:32.53 | scanline | ibot: explode is <action> explodes into a happy puff of smoke |
19:32.55 | | ACTION blows up is <action> explodes into a happy puff of smoke with bombs |
19:32.55 | DevGirl | ibot: blows up |
19:32.56 | | DevGirl: I give up, what is it? |
19:33.06 | DevGirl | :P |
19:33.10 | scanline | ibot: explode |
19:33.10 | | scanline: bugger all, i dunno |
19:33.14 | scanline | grr |
19:33.32 | Dentoid | ibot: I fart in your general direction. |
19:33.32 | | Dentoid: KCI error, or a problem with the Keyboard-Chair Interface. |
19:33.35 | lalo[work] | ibot: explode Dentoid |
19:33.37 | | ACTION blows up Dentoid with bombs |
19:33.40 | *** join/#picogui PicoBot (~PicoBot@ivan.client.mscd.edu) |
19:33.44 | lalo[work] | ibot: explode PicoBot |
19:33.46 | | ACTION blows up PicoBot with bombs |
19:33.46 | DevGirl | PicoBot!!!!! |
19:33.51 | lalo[work] | unfun. |
19:33.55 | Dentoid | ibot: insult lalo[work] |
19:33.55 | DevGirl | PicoBot: kiss |
19:33.57 | gonkulator | PicoBot: explode |
19:34.00 | Dentoid | Hmm, stupid bot. |
19:34.10 | DevGirl | :) |
19:34.18 | DevGirl | oh! what a happy life ^^ |
19:34.24 | lalo[work] | PicoBot: explode twice |
19:34.36 | gonkulator | ok, I don't have PicoBot joining all the channels it once did |
19:34.54 | gonkulator | but PicoBot is still invitable into them |
19:34.57 | scanline | I think it was in #tacobeam, #picogui, and #openembedded |
19:35.06 | lalo[work] | #agenda |
19:35.07 | gonkulator | way more than that |
19:35.09 | scanline | ah, right |
19:35.15 | gonkulator | #wonkyyokel |
19:35.16 | gonkulator | #pony |
19:35.20 | gonkulator | #aof |
19:35.26 | lalo[work] | eep |
19:35.29 | Dentoid | #eof |
19:35.35 | Dentoid | #aol |
19:35.37 | gonkulator | a few other places :) |
19:35.40 | Dentoid | #lol |
19:35.48 | scanline | gonkulator: just go ahead and make it autojoin all of them |
19:35.51 | scanline | might as well.. |
19:35.57 | gonkulator | alright |
19:36.55 | *** join/#picogui PicoBot (~PicoBot@ivan.client.mscd.edu) |
19:40.33 | *** join/#picogui Oktal (~mat@pc1-rdng3-5-cust199.winn.cable.ntl.com) |
19:46.07 | lalo[work] | scanline: nifty has a terminal too ;-) |
19:46.12 | *** join/#picogui gonkulator (~brandon@aden2-241-dhcp.resnet.colorado.edu) |
19:46.56 | scanline | oh? |
19:47.35 | *** join/#picogui gonkulator (~brandon@aden2-241-dhcp.resnet.colorado.edu) |
19:47.39 | gonkulator | whoa |
19:47.41 | lalo[work] | it's not obvious in any way yet, as it still needs tweaking |
19:47.46 | gonkulator | this is totally new |
19:47.50 | lalo[work] | but you can do: frame.open(Subprocess()) |
19:48.35 | scanline | cute |
19:48.46 | *** join/#picogui prae (~prae@sherpadown.net) |
19:48.47 | scanline | the scrolling's kind of not workful though |
19:49.00 | scanline | ah, I had to resize it before it worked |
19:49.05 | scanline | hmm |
19:49.41 | scanline | there's a lot of latency |
19:50.26 | *** join/#picogui prae (~prae@sherpadown.net) |
19:51.28 | scanline | lalo[work]: nifty's terminal is already way better than emacs' :) |
19:51.29 | lalo[work] | yup |
19:51.38 | lalo[work] | that's the main thing that needs tweaking |
19:51.44 | lalo[work] | right, it's better than emacs :-) |
19:51.48 | lalo[work] | it's a real vt102 |
19:52.14 | lalo[work] | I dunno how to fix the latency, perhaps lowering the idle_delay helps |
19:52.19 | scanline | well.. |
19:52.23 | lalo[work] | that's what prot did on epterm |
19:52.37 | scanline | cli_python needs a real way to add the subprocess's file descriptor to a select() or poll() |
19:52.54 | scanline | both epterm and nifty are way behind pterm in speed |
20:02.28 | gonkulator | or vaccumes |
20:02.31 | gonkulator | or vacummes |
20:03.19 | file | gonkulator: GENTOO FOREVER |
20:04.18 | *** join/#picogui DevGirl (maisa@200.203.12.232) |
20:04.56 | DevGirl | i'm back! |
20:05.08 | scanline | :) |
20:05.31 | scanline | hehe |
20:06.57 | lalo[work] | there can't be a select() or poll() |
20:07.07 | scanline | why? |
20:07.14 | lalo[work] | remember, you took back the feature where a request after WAIT stops the event wait |
20:07.39 | lalo[work] | so when I switched cli_python to a poll() loop, it started receiving duplicate events |
20:07.43 | scanline | I did? |
20:07.57 | scanline | cli_c still uses select()... |
20:08.08 | lalo[work] | you did... at least this is noted in the protocol doc |
20:08.25 | lalo[work] | what we could do IMO is add a request that waits with a timeout |
20:08.33 | scanline | but timeouts are no good |
20:08.34 | lalo[work] | btw: how do I update the online protocol doc? |
20:08.37 | scanline | we want immediate response |
20:08.54 | scanline | lalo[work]: edit the SGML source in the 'doc' directory in CVS, make, then upload the resulting HTML |
20:09.03 | lalo[work] | ah. |
20:09.04 | scanline | in fact I think the makefile might have an scp command in it |
20:09.07 | lalo[work] | I thought it was automated :-) |
20:09.12 | scanline | oh |
20:09.13 | scanline | maybe it is |
20:09.52 | scanline | ah, yep.. I have the same script doing the protocol doc and the cli_c doxygen stuff |
20:09.55 | lalo[work] | if it's not automated we can move it to the new site |
20:10.10 | scanline | it's scp'ing to sourceforge |
20:11.41 | lalo[work] | I think a timeout would work |
20:15.53 | lalo[work] | if you give the same timeout to WAIT and your select/poll, then it would be safe |
20:16.07 | lalo[work] | you know that if the server didn't send a response, then it won't |
20:33.26 | scanline | but that's missing the point |
20:33.34 | scanline | if you have to poll with a timeout, there will be a lot of latency |
20:33.54 | scanline | cli_python could do the same thing cli_c does- it avoids that race condition |
20:35.05 | lalo[work] | no |
20:35.08 | lalo[work] | it doesn't |
20:35.12 | scanline | hm? |
20:35.30 | lalo[work] | if the extra fd has data to read, what happens? |
20:35.49 | scanline | the select() exits immediately instead of timing out |
20:35.50 | lalo[work] | ah, ok, I can see a way to write it so that it doesn't have the race |
20:35.59 | scanline | cli_c explains what it does pretty well |
20:36.06 | scanline | read cli_c/src/netcore.c starting at line 127 |
20:36.11 | lalo[work] | then when the handler returns, it goes back to the select() without posting a new WAIT, right? |
20:36.15 | lalo[work] | that would work |
20:36.18 | lalo[work] | hmm |
20:36.18 | scanline | no |
20:36.27 | lalo[work] | except... what happens if the handler posts a request? |
20:36.29 | scanline | that doesn't necessarily work if the handler makes an API call |
20:36.30 | scanline | right |
20:36.42 | scanline | the client has to make sure it's off the waiting list and handled all incoming events |
20:36.56 | scanline | look at how cli_c does it, netcore.c:127 :) |
20:37.53 | lalo[work] | perhaps you could do that in cli_python? |
20:38.10 | lalo[work] | you'll find skeleton code in Server.py and network.py that does mostly this but not exactly |
20:38.13 | scanline | ok |
20:38.23 | lalo[work] | what I'm missing is a way to signal the server to take us out of the wait list |
20:38.46 | scanline | cli_c sends a 'ping' request, but any request should take it off the waiting list |
20:38.54 | lalo[work] | ping |
20:38.59 | lalo[work] | I forgot that exists :-D |
20:39.11 | lalo[work] | scratch network.py |
20:39.15 | lalo[work] | I meant Server and Application |
20:39.26 | *** join/#picogui Soopaman (~soopaman@h24-66-55-163.wp.shawcable.net) |
20:39.28 | Soopaman | hello |
20:39.37 | scanline | "IMPORTANT NOTE: The behaviour of this request will change in a future implementation of this protocol, the client should for now restrict itself from sending any other request after a PGREQ_MKWAIT is sent." <-- I don't think this is necessarily valid |
20:39.42 | lalo[work] | the server object already has a poll object |
20:39.47 | Soopaman | hello hello |
20:39.49 | lalo[work] | then remove it ;-) |
20:39.50 | scanline | hi Soopaman |
20:39.54 | scanline | lalo[work]: will do |
20:40.00 | lalo[work] | it scared the hell out of me |
20:40.10 | lalo[work] | hi Soopa |
20:40.32 | Soopaman | hi guys, what is up/new? |
20:40.36 | lalo[work] | ok, so you can add a method to Server to add something to the poll object, and then add a relay method to Application that calls this |
20:40.43 | scanline | Soopaman: Nifty, terminal, epterm, bugfixes... |
20:40.59 | lalo[work] | the poll() loop will have to change, it's not expecting poll to return anything but the socket |
20:46.43 | Soopaman | scanline, can you freely change the screen orientation (rotation) |
20:46.56 | scanline | yes |
20:47.51 | Soopaman | cool |
20:47.59 | Soopaman | on the fly? |
20:48.53 | lalo[work] | btw |
20:49.09 | lalo[work] | if you change the rotation by client request, does this go away when the client quits? |
20:51.27 | scanline | lalo[work]: no, it's a video mode flag |
20:52.10 | lalo[work] | ok |
20:52.23 | lalo[work] | what about loading and unloading themes? |
20:52.39 | lalo[work] | I figure I could read the protocol doc but it's easy to ask you ;-) |
20:52.43 | scanline | themes are objects owned by the client, so they do disappear when the app dies |
20:53.03 | scanline | I was planning on having a theme manager app that would own the themes, and provide a UI for inserting/deleting/rearranging them |
20:53.16 | lalo[work] | yup |
20:53.19 | lalo[work] | that would be useful |
20:53.20 | file | well it looks as though I'll be able to migrate over some stuff from Xandros into Gentoo... |
20:53.22 | scanline | it could read the "Theme tags" whenever those get implemented, to provide descriptions and screenshots for the themes |
20:53.24 | scanline | brb |
20:54.27 | lalo[work] | themeindex could be made into this app if it was able to hide its window |
20:55.32 | *** join/#picogui fukss (FKMF@202.88.190.103) |
20:55.43 | *** part/#picogui fukss (FKMF@202.88.190.103) |
21:20.16 | *** join/#picogui decaff (khedspet@3Cust236.tnt19.tpa2.da.uu.net) |
21:28.00 | Oktal | anyone happen to have a bit of autoconf code for finding libpgserver? |
21:29.26 | *** part/#picogui decaff (khedspet@3Cust236.tnt19.tpa2.da.uu.net) |
21:30.24 | lalo[work] | nope |
21:30.35 | lalo[work] | perhaps there is something like that in the apps module |
21:30.46 | Oktal | mm |
21:31.05 | scanline | doubt it |
21:31.25 | scanline | though.. isn't there an autoconf macro that tests for a function in a library? |
21:31.42 | Oktal | i have no clue about autoconf |
21:31.45 | scanline | yeah, AC_CHECK_LIB |
21:31.53 | scanline | you should be able to do something like: |
21:32.32 | scanline | <PROTECTED> |
21:34.44 | Oktal | looking at CS's configure.ac it seems it defines its own macro for this |
21:35.00 | scanline | mmkay |
21:35.14 | scanline | I haven't gone real deep into auto*, since most of it really disgusts me :) |
21:35.27 | scanline | I'll be replacing auto* in picogui with something python-based in the nearly close future |
21:35.34 | Oktal | me too |
21:35.35 | Oktal | nice |
21:35.58 | Oktal | CS will probably be converting to the Jam makefile system |
21:36.10 | scanline | I've looked at Jam a little.. but I'm leaning toward SCons |
21:36.27 | scanline | SCons looks to be the most easily extensible, which will help us with all the custom code we'll need for configuration |
21:37.39 | Oktal | where are pgserver lib and includes installed by default? |
21:39.44 | scanline | /usr/local/lib and /usr/local/include by default |
21:39.49 | scanline | lalo[work]: screenshot resizing? :) |
21:41.17 | Oktal | ok thanks |
21:42.24 | TD | scanline: are you sure that's a good idea? |
21:42.35 | scanline | which idea? |
21:42.39 | TD | sure autoconf makes me want to puke too, but from what I saw of SCons python isn't the best language for a build system |
21:42.47 | scanline | heh |
21:42.54 | scanline | what makes you say that? |
21:42.54 | TD | it seemed to be rather verbose |
21:43.09 | scanline | well, picogui needs a lot of customization, so verbosity might not be a bad thing for us |
21:43.13 | TD | as opposed to bash, which for all its uglyness at least can do greps, regexs, cps etc easily |
21:43.18 | TD | hmm, true |
21:43.27 | scanline | and python can do regexes and such pretty easily |
21:44.33 | TD | the syntax seems to have changed since i last saw it |
21:45.13 | TD | strange |
21:45.14 | TD | oh well |
21:45.19 | TD | can it do autoconf style probes/checks? |
21:46.52 | scanline | no, that's one of the major gaps in SCons' feature set |
21:47.16 | scanline | but I expect I'll be writing that myself no matter what system picogui moves to, might as well do it in python |
21:47.42 | lalo[work] | scanline: isn't screenshot resizing already there? |
21:47.50 | TD | well, good luck |
21:47.50 | scanline | lalo[work]: yeah, but it looks bad :) |
21:47.57 | lalo[work] | does it still? |
21:47.59 | scanline | TD: thanks |
21:48.09 | lalo[work] | I thought George had removed PIL already |
21:48.20 | scanline | lalo[work]: yeah, it's not interpolating |
21:48.31 | lalo[work] | PIL doesn't interpolate |
21:48.48 | scanline | hmm.. whatever we were using at first tid |
21:48.50 | scanline | err, did |
21:49.00 | lalo[work] | if he removes one line from Photo.py so that it doesn't use PIL, it will revert to ImageMagick which interpolates |
21:49.22 | scanline | hmm |
21:52.04 | lalo[work] | perhaps I could make this change to CMFPhoto's CVS and ask him to update ;-) |
21:52.41 | lalo[work] | what I did was solely for us... I made it easier to post a news item |
21:52.52 | scanline | ah, that's good |
21:53.05 | lalo[work] | by adding a "add new" button to the main /News page which only appears if you have the permission to do it |
21:53.12 | scanline | cool |
21:53.23 | lalo[work] | and all that because I wanted to add a news item with the roadmap |
22:39.38 | *** join/#picogui Oktal_ (mat@pc1-rdng3-5-cust199.winn.cable.ntl.com) |
22:56.06 | Soopaman | http://tv.yahoo.com/news/wwn/20030319/104808600007.html |
22:56.08 | Soopaman | sweeet |
22:56.44 | DevGirl | yawn |