--- Log opened Thu May 14 00:00:32 2009
00:54 < mae_> poke
00:56 < mae_> damn this elusive fileServe madness
01:00 < abuiles> hey!
01:00 < mae_> hey so, this build auto thing, i'm hearing lots of problems
01:00 < mae_> that sucks
01:00 < mae_> lets see if i can fix it.
01:01 < abuiles> Yeah,,, I'm not really using it, I was just "testing"
01:01 < mae_> ok so
01:01 < mae_> what were your findings
01:01 < mae_> lets troubleshoot
01:02 < mae_> you said that it kept overwriting the binary
01:02 < mae_> even though there were no changes?
01:02 < abuiles> It does that, like 3 times, then..
01:03 < mae_> ok stop right there
01:03 < mae_> 3 times without changing the files at all?
01:03 < mae_> please paste the exact command
01:04 < abuiles> first, which binPath should I point ?  the one in .cabal ? or the one in /dist ?
01:05 < mae_> what are you using to build it?
01:05 < mae_> bin/build.bat?
01:05 < mae_> what is your platform?
01:05 < abuiles> linux mint.
01:05 < mae_> ok
01:05 < mae_> so your using bin\build.sh?
01:07 < abuiles> have you checked the mailing list ? I wrote there the results of the new test which I did today, with "cabal configure & cabal install "
01:07 < mae_> ok
01:08 < mae_> so your using the cabal script
01:08 < mae_> ok
01:08 < mae_> so what i do is
01:09 < mae_> oh wait
01:09 < mae_> i think i know what your prob is
01:09 < mae_> cabal install will overwrite the binary for the install step no matter what
01:09 < abuiles> ?
01:09 < mae_> so yeah what you do is
01:13 < abuiles> mae_:  which <binPath> have you used in your tests ?
01:13 < mae_> sec
01:15 < mae_> for the guestbook
01:15 < mae_> http://msdn.microsoft.com/en-us/library/ms740565.aspx
01:15 < mae_> arg
01:16 < mae_> happstack build auto "cabal configure && cabal build" dist/build/guestbook-server/guestbook-server
01:16 < mae_> for the guestbook
01:16 < mae_> (cabal puts the binary in dist/build after building)
01:16 < TomMD> Is there a preliminary document on the 0.2 -> 0.3 API changes?
01:17 < mae_> but it only links this exe (hence changing the last modify date) IF there was actual changes
01:18 < mae_> TomMD: nothing like that yet, but as far as backwards compatibility goes, I don't believe this should change much, at this point, it is hard for me to remember everything that has changed, but AH you could take a gander at http://patch-tag.com/r/happstack/snapshot/current/content/pretty/happstack/CHANGELOG
01:18 < abuiles> So, did you see, than I tried with that command as well, and it didn't work.? It wasn't still detecting any changes in the build.
01:19 < mae_> abuiles: indeed, i think i am having the same problem ATM :)
01:20 < abuiles> I was doing something like "putStrLn "inside killLopp" and It just appeared once..
01:20 < abuiles> I think the problem is around addTimeOut
01:21 < mae_> yeah ok
01:21 < mae_> like it never tried to keep building
01:22 < mae_> the buildKillLoop is not working..
01:22 < abuiles> right
01:23 < abuiles> other thing, is that when I kill with ctrl-c the process continues executing..... I have to do "kill"
01:25 < mae_> yeah
01:25 < mae_> well
01:25 < mae_> it kills it violently if it does kill it
01:25 < mae_> it doesn't pass through a graceful sigterm to the process
01:25 < mae_> we need to catch the signal so as to bypass the normal order of things
01:26 < mae_> or hey if any haskell gurus know how to make forkIO'd threads automatically get a kill signal when the main loop gets killed...
01:28 < mae_> ah i see
01:28 < mae_> so the "main" loop as i am calling it, dies, but then the spawnLoop keeps the process alive
01:46 < mae_> this autobuilding tool sucks right now, i definitely need to fix this
01:51 < mae_> solidsnack: hola
01:52 < solidsnack> mae_: Hello.
01:52 < solidsnack> How do you do?
01:52 < mae_> allright
01:52 < mae_> too much stuff, not enough time
01:52 < mae_> want to hack on haskell all day
01:52 < mae_> but all i do all day is windows
01:53 < mae_> and now diapers around the corner
01:53 < mae_> : )
01:53 < solidsnack> windows?
01:53 < solidsnack> You are a Windows dev?
01:53 < mae_> no
01:53 < mae_> even better
01:54 < mae_> sysadmin
01:54 < mae_> : )
01:54 < solidsnack> Oh, wow. That's heavy.
01:54 < mae_> definitely
01:54 < mae_> i recently have been writing a win32 cabal package that wraps up various psapi and kernel32 dll stuff for process management
01:54 < mae_> so i can kill processes on our citrix terminal servers
01:55 < mae_> yes, i'm a glutton for punishment
01:55 < mae_> i wrote a sysadmin script that would have taken 10 minutes in a .bat file
01:55 < mae_> and instead spent a week in haskell
01:55 < mae_> (and building plumbing)
01:56 < mae_> paulvisschers: hi
01:58 < paulvisschers> Hey
02:01 < solidsnack> mae_: Well, that's actually a good thing.
02:02 < mae_> lo
02:02 < mae_> lol
02:02 < mae_> ok i think i found the problem!
02:02 < mae_> addTimeout does not block!
02:03 < mae_> bummer
02:03 < mae_> solidsnack: in what sense is it a good thing?
02:10 < abuiles> great.
02:12 < mae_> or...
02:12 < abuiles> mae_:s o, how could you fix that ?
02:12 < mae_> waitForProcess is blocking too much
02:12 < mae_> it should be respawning in 5 secs
02:13 < mae_> hmm
02:15 < abuiles> is "waitForProcess" really important ? I mean, because if respawning so It wouldn't reach it.
02:20 < mae_> yeah it is because
02:20 < mae_> otherwise we would keep launching processes over and over
02:21 < mae_> waitForProcess waits for the process to die, for whatever reason (runtime error, or the buildKillLoop got it, or out of memory, anything really)
02:21 < mae_> it ensures that we are only running one instance of the executable at any time
02:21 < mae_> addTimeOut is nonblocking
02:43 < solidsnack> solidsnack: I mean, what your throwing together will help a lot of folks.
02:45 < solidsnack> s/your/you're/
02:52 < mae_> solidsnack: well i'm duplicating alot of what is in System.Win32
02:52 < mae_> but that stuff isn't typesafe.
02:52 < mae_> and is missing alot of stuff i need
02:52 < mae_> + i'm using c2hs
02:53 < mae_> found several bugs related to cabal + c2hs on windows
02:53 < mae_> talked with dcoutts about them
09:44 < gour> hi, not 100% connected with happs, but does anyone know if turbinado project is still alive (no updates since march) and/or does fcgi support in happs makes it kind of obsolete?
10:25 < cocon> how can we events for comet-style pending connections be triggered/listened to?
10:42 < stepcut`> cocon: comet can mean a lot of things..
10:45 < stepcut> don't most people just use polling via XMLHttpRequest now ?
--- Log closed Fri May 15 00:00:35 2009