IRC channel logs

<civodul>stis: impressive work!

<ijp>civodul: how do you handle working around guile bugs in guix?

<civodul>do you really want to know? :-)

<stis>civodul. I think I release it, it's good to have a tag and not develop forever.

<civodul>yeah :-)

<civodul>ijp: Guix still has to support 2.0.5

<civodul>so it has a copy of a subset of one or two web modules

<civodul>and a copy of SRFI-37

<civodul>i think these are the main things

<ijp>so straight copying, no fancy hacks

<civodul>well, for the web stuff it's terrible

<civodul> http://git.savannah.gnu.org/cgit/guix.git/tree/guix/http-client.scm#n56

<stis>ok I released a new version, night hackers!

<cky>mark_weaver: Congrats on your progress on Google Code Jam!

<cky>Currently I only have A-small and B-small solved, but I've only been working on it for 1.5 hours (and I spent much of the time on problem A reacquainting myself with Rackety stuff).

<cky>Am trying to decide on whether it's best to focus the remaining 2.5 hours on B-large or C-small. :-P

<davexunit>ooh google code jam

<davexunit>didn't know about it

<cky>davexunit: You still have 2 hours left.

<cky>You can easily wing it in that time.

<cky>You might not solve all 4 problems in 2 hours, but if you successfully solve 2, you'll get through.

<davexunit>cky: I'll be too distracted by other things going on around me.

<davexunit>it looks fun, though.

<cky>Problems A and B are really easy. I promise you.

<davexunit> I'm too busy writing particle simulations :P

<cky>Oh dear. :-P

<mark_weaver>cky: thanks! we'll see whether I got the large problems correct. *crosses fingers* :)

<mark_weaver>cky: looks like you got a late start.

<mark_weaver>good luck! (it's better if I don't distract you)

<cky>mark_weaver: :-D I have 1 hour to formulate a solution for problem C, though it's been stewing in my brain the whole day. Here's hoping!

<mark_weaver>that was a tough one.

<cky>mark_weaver: But yes, late start for me indeed.

<cky>Is it? Uh oh.

<mark_weaver>D was easier, fwiw.

<cky>Huh.

<cky>Okay, lemme try. :-)

<mark_weaver>although if you've been thinking longer about C, maybe that will make up for it, dunno!

<mark_weaver>good luck!

<cky>Thanks!

<mark_weaver>(I'd still recommend doing C at some point, even if not before the qualification round ends, because it's an interesting problem)

<ijp>mark_weaver: have you ever read "Adding Delimited and Composable Control to a Production Programming Environment"?

<mark_weaver>ijp: no, I haven't heard of that one.

*mark_weaver looks it up

<ijp>I'm impressed by the care it takes to make everything work right

<mark_weaver>I'll add it to my reading list.

<mark_weaver>ijp: Thanks for bringing that to my attention. Hopefully Andy already read it and incorporated its lessons into Guile.

<ijp>nalaginrut brought to my attention a problematic interaction between abort and reset, and I've been looking to see if there was a way to fix it

<ijp>note "reset" not "prompt"

<ijp>I *think* the way racket handles it is that their low level version of abort returns a thunk, whereas ours does not

<ijp>so the extra work we do in reset isn't necessary

<cky>mark_weaver: Congrats on your perfect score.

<ijp>3 years in a row, isn't it?

<mark_weaver>thanks!

<mark_weaver>I don't remember if I got a perfect score last year. I think maybe not.

*mark_weaver looks

<mark_weaver>one thing I'm pleased about this year is that my solutions were all purely-functional and yet still fast enough to run in just a few seconds even on my YeeLoong.

<mark_weaver>(and of course, all written in Guile)

<mark_weaver>I do recommend trying to do problem C before looking at the answer.

<mark_weaver>well, if you're bored, anyway :)

<mark_weaver>I didn't get a perfect score last year. The last problem stumped me.

<mark_weaver>(Treasure)

<cky>mark_weaver: Wait, what, your problem C is purely-functional?!

<cky>mark_weaver: I actually used a vector for that.

<mark_weaver>yep :)

<mark_weaver>my guess is that my solution works in a much different way than yours

<cky>Um, yeah. ;-)

*ijp surreptitiously slides a url to pfds underneath cky's nose

<mark_weaver>when I have enough time (and even sometimes when not), I tend to just try to solve the large problem, and not even think about the small one.

<cky>I do like my 13-line solution to problem A. (All my solutions this time are in Racket.)

<cky>ijp: Does Scheme have pure-functional vectors?

<mark_weaver>I didn't use vectors at all.

<ijp>cky: I have semi-functional vectors, and sequences based on trees

<cky>mark_weaver: *nods*

<cky>By semi-functional, do you mean copy-on-write? ;-)

<ijp>the difference between our primitive continuations and rackets, is that racket has a primitive way of capturing the current delimited continuation without abort

<ijp>cky: no

<ijp>cky: it uses mutation under the hood

<ijp>it doesn't do any copying

<ijp> https://github.com/ijp/fectors

<ijp>anyway, the above difference means I don't think I can fix the interaction with the current set of operators

<ijp>because to get the continuation I need to abort, and have the call-with-prompt fix it up

<ijp>which means a special handler, and anyone else using the default prompt is going to get that handler

<ijp>I'll discuss with Andy tomorrow, and maybe a report

***linas_ is now known as linas

<cky>mark_weaver, ijp: https://gist.github.com/cky/10566479 <-- my GCJ code if you're curious

<mark_weaver>all of our solutions are now available from google.

<cky>I know, but each entry requires a separate download, I have to unzip each, etc.

<cky>so I provided the Gist as a convenience for people who just want to glance.

<mark_weaver>cky: nice concise solutions :)

<cky>:-D

***alezost`` is now known as alezost

***Fuuzetsu is now known as Guest17487

<stis>hej guilers!

***Guest17487 is now known as Fuuzetsu

<e2x>When I type (display "\\n") into the repl, it displays a newline as expected. If I have a script with (display (cadr (command-line))) and type ./script "\\n", it displays \\n. What is an easy way to get the latter case to display an actual newline?

***linas_ is now known as linas

<ijp>is "\\n" a newline in the shell?

<ijp>hmm, no

<ijp>e2x: the problem then is that guile sees a string with two characters, whereas in guile source \\n is treated as one

<ijp>I can't think of any easy way to do it from string escapes, though I know there is for format escapes

<ijp>but, you can get the shell to handle C style escapes (or at least bash can)

<ijp>./script $'foo\\n'

<ArneBab>my first FFI test: http://paste.lisp.org/display/142024

<cky>ArneBab: Did it work?

<ArneBab>yepp - feel free to try it ☺

<ArneBab>it gives the version of the netcdf4 library

<cky>Nice. :-)