Back to Home / #mythtv / 2002 / 10 / Prev Day | Next Day
#mythtv IRC Logs for 2002-10-07

10:58--> Youare now talking on #mythtv
12:49-!-You are now known as mdz_
20:32-!-gtaylor [] has joined #mythtv
20:32<gtaylor>hiya. seen my little two-player mystery report?
20:33<Chutt>nope, not yet
20:33<Chutt>just send it?
20:33<gtaylor>maybe ten minutes ago. maybe the list doesn't like me again?
20:33<Chutt>i haven't gotten an admin mail
20:33<Chutt>one of the servers my email goes through is a tad slow right now
20:34<Chutt>oh, actually
20:34<gtaylor>oh: Connection timed out with
20:34<Chutt>i was just going to say that =)
20:34<Chutt>his network seems to be having problems at the moment
20:35<Chutt>so what'd you say in the email?
20:35<gtaylor>ok, view thee
20:36<gtaylor>i'm examining state transitions in tv.cpp wiht an eye toward extra player starts when recording starts/stops, but it doesn't look wrong.
20:36<Chutt>each distinct thread that needs to talk to the database has its own connection
20:37<Chutt>so that's 3, and one X connection make 4
20:37<Chutt>not sure about 2 fds on the playback, though
20:37<gtaylor>ah, well, that's the sockets then. socekt leaks don't concern me so much ;)
20:38<gtaylor>that's what I thought ;)
20:38<Chutt>i've never seen anything like that
20:38<gtaylor>is there exactly one thread telling Tv to "do this" "do that"?
20:39<Chutt>thread break down:
20:39<gtaylor>could be a race between them?
20:39<Chutt>there very well could be a race condition somewhere
20:39<Chutt>but, hmm
20:39<gtaylor>the api appears to be "do this" followed by "state change" call, so it's arguably race-prone if called from two places.
20:40<Chutt>but that'd be more of it not doing something
20:40<Chutt>i suppose i could lock that all down
20:40<Chutt>but i don't think that's it
20:41<gtaylor>yeah, i don't see any path from recording A watching Z -> recording B watching Z that would get you a bonus Z
20:41<Chutt>if you can reproduce that high load average while running in gdb
20:41<Chutt>if you ctl-c it in gdb
20:41<Chutt>and do an 'info threads'
20:42<Chutt>should at least be able to tell which threads are doing what
20:42<Chutt>if you've noted the high cpu% pids
20:43<gtaylor>hmm. why is the deafult build symbol-free?
20:43<Chutt>since it needs to build without MMX and all that turned on for completely proper gdb operation
20:44<gtaylor>oh, gdb doen't mmx? figures...
20:44<Chutt>it was messing up for me, at least
20:45<gtaylor>ok, well, I won't get a chance to gdb the thing until later. and I haven't got that exact version's source anymore, so it'll just be with cvs head. we'll see what we see
20:45<Chutt>later =)
20:45-!-gtaylor [] has quit ["Leaving"]
20:51-!-gtaylor [] has joined #mythtv
20:52<gtaylor>c: here's anothe rtheory: perhaps there's a spurious video preview thread getting started from the playbackbox somehow.
21:10-!-gtaylor [] has quit ["Leaving"]
21:24<vektor>Chutt: Around?
21:30<vektor>When you wake up let me know.
21:31<Chutt>i'm around
21:31<vektor>Ok, I'm playing with text rendering a bit.
21:31<vektor>I noticed that in ttfont.c you set the chroma to b/w. Why is that again?
21:31<vektor>Like, why aren't you compositing colour text?
21:31<Chutt>because i'm only doing b/w text
21:32<vektor>Ah ok.
21:32<vektor>Next question, do you know what the font size really means?
21:32<vektor>Is it height in pixels or what?
21:32<vektor>Like, I know nothing about what '24 pt font' really means.
21:32* vektorstupid.
21:32<Chutt>it's either points or pixels
21:32<Chutt>not really sure which in there
21:32<vektor>Yeah but if it's points what does that means?
21:32<vektor>erm, mean? :)
21:33<vektor>bah I'll figure that out.
21:33<vektor>Anyway, my text compositor does colour and I have nice alpha fades.
21:34<vektor>It's cool. When I change channels in my TV app it shows the channel number and fades it out quickly. ;-)
21:34<vektor>I'm currently trying to figure out how to get better borders cheaply.
21:34<Chutt>i think pointsize is = pixel size on 75 dpi screens
21:34<Chutt>or something around that
21:34<vektor>Well that's useless.
21:35<vektor>(for this purpose) :)
21:35<vektor>I'm unimpressed with the alpha values I get from freetype.
21:35<vektor>Do you know if that's just crappy fonts, or what?
21:35<vektor>Like, I'm thinking we should supersample and do a better job.
21:35<Chutt>have you tried the ms fonts?
21:36<vektor>Dunno. I started with 'helr.ttf' and now I'm using Verdana which looks nicer but I haven't checked the alpha values.
21:36<vektor>It's only like 2 bit alpha though which is surprising.
21:36<Chutt>the code there _is_ messing with the alpha
21:36<vektor>It is?
21:37<Chutt>that whole alpha_lut, bounded_palette thing
21:37<vektor>Yeah, I export that.
21:37<vektor>But it looks like the font can only give you five values.
21:37<vektor>And so you only get like 5 values of alpha.
21:37<vektor>Instead of like the full 0-255 range.
21:38<vektor>oh waitasec
21:38<vektor>how is 'bounded_palette' used?
21:38<Chutt>to make it only have 5 values of alpha =)
21:38<vektor>Oh damnit.
21:39<vektor>Well how do I get more? :)
21:40* vektortries something random
21:41<vektor>argh, didn't work :)
21:46<vektor>argh this is annoying
21:47<Chutt>wish i could help ya, but it's really been a few years since i touched that font stuff
21:47<Chutt>and i didn't even write most of it in the first place
21:47<vektor>well did you write this bounded_palette stuff?
21:47<vektor>i'm just wondering
21:47<vektor>do you know what 'iread' and 'ioff' might possibly mean?
21:47<vektor>like, do you know if maybe tt fonts have 10 quantization levels?
21:47<Chutt>i think that's what it is
21:48<Chutt>the freetype docs are fairly good
21:48<vektor>ok thanks.
21:48<vektor>I'll check them out.
23:11-!-mdz-away is now known as mdz