<didi>Does guile read my ~/.guile when I run programs from the command line like `$ guile foo.scm'? <ijp>I think so, unless you use -s <ijp>no, only during interactive use <didi>ijp: Hum. It doesn't look like it does. I add a directory to the load path inside my ~/.guile and I can `,use' it from the REPL, but `guile foo.scm' reports there is no code for the module. <didi>Hum, I must use another strategy for adding directories to the load path. ***turbofai` is now known as turbofail`
<ijp>GUILE_LOAD_PATH or -L <didi>I guess I'll go to the env variable so then geiser can take advantage of it too. ***heroux_ is now known as heroux
***sneek_ is now known as sneek
<didi>If I have a library that uses GOOPS and I want users to instantiate objects from a class <foo>, what do I #:export? <ijp><foo> and any new generics you've written <didi>ijp: Nice. Will users have to use-module (oop goops)? <ijp>unless you also export define-method etc. <didi>Does guile move structures created with `make-c-struct' around? <didi>Meaning, do I have to worry about (pointer-address (make-c-struct ...)) changing? <mark_weaver>even if we moved to a moving GC, I'm sure that we'd arrange to make sure that anything created by make-c-struct would never move. <didi>mark_weaver: Oh, nice. Thank you. ***turbofail` is now known as turbofail
<stis>how do I get the module a procedure is defined in? <taylanub>sneek: Later tell didi on my 2.0 GHz Core2Duo I can make a million references in 6 seconds for the fourth field of a struct within a vector within a struct, or 8 seconds if I reference the 7th field of said struct (the depth of the type (layers of vectors/structs) and, in case of structs, the position of the field, affect the referencing speed) <taylanub>Hrm, I should just put those results into the docs. <stis>nay there isn't such a binding, soI need to use procedure-proerties and roll my own binding <unknown_lamer>wingo: I created a vectorizing abstraction based on cut, once I am done slogging through hcoop crap (I guess I have to rekey an afs cell, ugh) I am totally going to port it to work atop compost <wingo>stis: it's quite tricky in master <wingo>well, stable-2.0 as well for that matter <wingo>if a function doesn't reference any toplevel bindings, is it defined in a module at all? <stis>yep, I just decided to roll my own system useing procedure-properties. ***mario-go` is now known as mario-goulart
***ota_ is now known as vojjta
***vojjta is now known as ota_
***ota_ is now known as ota
<wingo>oooh, i might have found the segfault <wingo>that's a pretty weird one :) <stis>crunch!! (the sound of a squached bug is sooo pleasing) <wingo>mark_weaver: did you see i removed the hard stack size limit in master? :) <wingo>we still need some nice way of controlling the soft limit <wingo>but probably we can default to not having a soft limit <wingo>as we don't have a limit for the heap <wingo>and install a limit in the repl <stis>Happy friday dsmith-work! ***wingo_ is now known as wingo
<wingo>are there any examples of libs that guile has incorporated into its distro for which that has been a bad idea? <wingo>statprof, sxml, etc -- all good in practice, it seems to me <civodul>wingo: i can't think of any bad example <civodul>we've only made smart choices, obviously <civodul>is it a bug that (string-locale<? "a" "B") => #t in en_US locale? <wingo>it's not a specified procedure, is it?