IRC channel logs


back to list of logs

<kyamashita>Anyone here use guile-wm with the latest Guile (2.1.5)?
<thomassgn>I made it into a list for some reason earlier, but it started as a string...
<dsmith>Ok if it's a list then don't wrap it in list.
<dsmith>So (apply system* r t o p (append files d))
<dsmith>thomassgn: Not sure thats the best way to do what you want. Haven't been doing much Schemeing lately, and answering questions helps me think.
<dsmith>thomassgn: If no one else here speaks up, try #scheme
<thomassgn>dsmith: Nice, closer than I've been so far :) But it gives me 'ERROR: In procedure length: Wrong type argument in position 1: ("/run/current-system/sw/bin/rsync" "-e lshg" "-zavh" "--info=progress2" "testfile" "tst" . "")'
<thomassgn>not sure if position 1 is the rsync command or "-e lshg"... or maybe it's complainig about the dot...
<thomassgn>not using 'length' anywhere though, guessing its called in the system* procedure...
<dsmith>thomassgn: That whole thing is position 1
<dsmith>I think.
<dsmith>thomassgn: Sorry, gotta go...
<thomassgn>dsmith: no worries, thanks a lot!
<nalaginrut>I'm trying to convince people to use Autoconf&Makefile for new project in our company. But seems peopl;
<nalaginrut>I'm trying to convince people to use Autoconf&Makefile for new project in our company. But seems people dislike the aging GNU building system anymore
<nalaginrut>I'm considering maybe we need a new one written in Guile and compatible with GNU building system...
<nalaginrut>alas, I can make all things done with handwritten Makefile in 20 minutes...oh well
<ArneBab_>nalaginrut: did you tell them about distcheck?
<ArneBab_>nalaginrut: maybe this can help:
<ArneBab_>nalaginrut: when I’m home this evening I can upload a pretty clean Makefile setup I’m using nowadays
<ArneBab_>including `make help` with targets and colored output.
<nalaginrut>ArneBab_: I never use distcheck, what's the usage? BTW, we just support GNU/Linux, so maybe it's unnecessary?
<nalaginrut>sorry, Debian/Linux
<civodul>hello nalaginrut!
<nalaginrut>civodul: heya
<civodul>nalaginrut: i'd suggest never publishing a tarball not produced by "make distcheck"
<civodul>it's really of great help
<nalaginrut>ok, I'll add TODO of Artanis ;-)
<ArneBab_>nalaginrut: distcheck builds a tarball, then unpacks it and checks whether the tests work for the program built entirely from the tarball.
<ArneBab_>so it tests whether what you publish actually works.
<nalaginrut>ArneBab_: thanks for explain!
<nalaginrut>it seems re-export the imported symbols doesn't work, but I can use export directly
<nalaginrut>alright, seems #:re-export must be above the #:export
<nalaginrut>still fail...
<nalaginrut>my bad, it's bug in my module organizing
***amz31 is now known as amz3
<dsmith>Happy Friday, Guilers!!
<OrangeShark>Happy Friday!
<paroneayea>I forgot to announce the new Guile release.
<paroneayea>btw there isn't a news post on the site still, will there be one or should I just point to the mailing list?
<civodul>speaking of which, wingo: what do you need to publish the announcement on the web site?
<paroneayea>ha :)
<civodul>yeah :)
<wingo>civodul: an updated README.MAINTAINERS in the cvs
<wingo>and some kind of README in guile-web or something
<paroneayea>should I wait to announce it till we have the web version then?
<paroneayea>the website does look so nice! :)
<civodul>wingo: updated README.MAINTAINERS
<civodul>for the news entry, just drop blog/foo.html in the Git checkout
<wingo>ok :) i am real busy at work and won't be able to get to it directly
<civodul>it should be pseudo-HTML, for use by Haunt's "HTML reader"
<davexunit>oh are the news entries just plain html files?
<davexunit>you guys are using haunt and I have no idea how it's all put together :)
<civodul>davexunit: i imported previous entries as sxml, but html seems ok for the new ones
<civodul>suggestions welcome though :-)
<paroneayea>civodul: I've been using Skribe for 8sync's site fwiw
<davexunit>is the HTML generated from something?
<wingo>if someone wants to put up a news item for the release, it is very welcome too :)
<civodul>wingo: i'm happy to help with haunt but i'd rather let you write the text
<paroneayea>though I guess civodul is probably also using skribe, I haven't looked at it :)
<davexunit>I use markdown for my blog
<wingo>i can throw up some text on a paste
<paroneayea>huh wow
<paroneayea>I also didn't realize that there was pre-emptive threading support in this new version of guile.
<paroneayea>that's pretty crazy
<davexunit>I still can't figure out how to use it :)
<davexunit>reading statprof only left me more confused, but apparently there is some SIGPROF signal that I can use.
<davexunit>confused about what is a UNIX signal and what is a user signal
<davexunit>api seems to conflate the two
<paroneayea>(I think "statistics professor" every time I use statprof)
<wingo>davexunit: see asyncs.test
<davexunit>wingo: thanks
<wingo>it has an example of using signals for preemption
<davexunit>wingo: "preemption via sigprof"?
<davexunit>will study this
<davexunit>ah, you can only specify that you want to interrupt after so many seconds
<davexunit>I was hoping to be able to interrupt at every possible opportunity.
<davexunit>maybe a time of 0 will do that...
<civodul>thanks wingo
<wingo>you need to use vm hooks i think if you want synchronous callbacks at every opportunity
<davexunit>I'll take a look
<civodul>news entry is (finally) online!
<davexunit>link 404s :(
<civodul>yeah it should vanish Real SOon
<davexunit>oh okay
<civodul>the beauty of upload-via-CVS
<davexunit>wingo: is (system vm traps) where I want to be looking?
<wingo>civodul: tx for help :) i will use that commit as a pattern in the future :)
<davexunit>wingo: thanks
<davexunit>trap-instructions-in-dynamic-extent looks like the right thing
<davexunit>but can I capture a continuation when this trap is triggered?
<davexunit>would like to say "you get N instructions then you will be interrupted and someone else will get a turn"
<paroneayea>wingo: so I guess a way to think about it is that suspendable-ports gives us clean cooperative behavior via being able to suspend on IO
<paroneayea>whereas pre-emptive threading allows the same thing, but suspending on CPU
<paroneayea>would that be right?
<wingo>davexunit: i think you can't capture a continuation
<davexunit>wingo: okay
<wingo>for fine-grained interrupt handling, threads are the right thing. you can say "you get 50us"
<wingo>i.e. have one thread that just interrupts your main thread
<davexunit>I think I've exhausted all the possibilities for implementing a Batlecode-like system.
<wingo>problem is, not all interrupt points are suitable for capturing continuations
<wingo>we need a prompt-resumable? predicate
<wingo>anyway mutually untrusting code competing for resources is a bit hard with guile for other reasons (memory usage, mainly)
<davexunit>yeah, I was only concerned about this one aspect.
<davexunit>maybe there is a better way, but it's how Battlecode ensures each player gets a fair amount of time to run their AI
<davexunit>for what I was thinking I am OK with assuming the code isn't malicious, I just want to be fair in how much processor time is allocated to each player.
***holomorp1 is now known as holomorph
<davexunit>paroneayea: not guile related, but this webgl rendering demo is funny and impressive
<davexunit>and something to strive for with regards to my own graphics programming adventures in guile
<davexunit>so many bunnies.
<paroneayea>davexunit: wow, works well even on my x200
<davexunit>paroneayea: nice!
<ft>Heh. I think using master with a project of mine just cut execution time down to a third of what it was before. :)
<ft>Kudos to all who made that possible! :)
<sneek>ACTION points the Dread Finger of Accusation straight at wingo
<ft>14 seconds versus 5.5 seconds. Not quite a third but close enough. :)
<OrangeShark>lol that an interesting response by sneek