IRC channel logs

<cmhobbs>is it possible to fire up a guile repl in emacs without opening multi-term first?

<cmhobbs>or some other terminal

<davexunit>cmhobbs: check out geiser.el

<davexunit>once installed, you can use M-x run-guile

<cmhobbs>also is there a concept of null or nil in guile? if so, what's the keyword?

<cmhobbs>i have geiser! thanks for the tip!

<davexunit>cmhobbs: typically we use false, #f

<davexunit>various data types have their own null value

<davexunit>like "" for strings and '() for lists

<davexunit>predicates string-null? and null? test for them, respectively

<wingo>meep meep

<civodul>hey Guilers!

<nalaginrut>heya

<nalaginrut>wingo: It's 3.12s now, while Python2 costs 2.4s. I don't think I can go any further without patching guile-core ;-)

<nalaginrut>wingo: here's the code https://gist.github.com/NalaGinrut/4418448e73b65af58ffa

<wingo>cool, will look

<wingo>civodul: slot-ref was a lot slower with the goops refactor; will work on that before any merge

<civodul>oh, ok

<civodul>even once it's inlined?

<dsmith-work>Thursday Greetings, Guilers

<davexunit>another day, another one-off guile script helping me do my job.

<davexunit>thanks guile!

<dsmith-work>davexunit: yey!

<civodul>guile: ../../libguile/ports.c:2527: scm_i_port_iconv_descriptors: Assertion `pti->encoding_mode == SCM_PORT_ENCODING_MODE_ICONV' failed.

<civodul>mark_weaver: does that ring a bell? ↑

<civodul>that's for a port whose encoding is #f

<mark_weaver>civodul: yes. the last time, it was because the pti structure was getting corrupted. memory corruption caused elsewhere.

<mark_weaver>I've forgotten what the source of the corruption turned out to be.

<civodul>do you remember if something was fixed in that area?

<civodul>oh ok

<civodul>it's 2.0.9.219-e1bb7

<mark_weaver>I vaguely recall a bytevector realloc bug that might have been the cause, but I don't remember if the link was clear.

<mark_weaver>commit 1baa2159307c34683e8ede54f38f65010fc594b0

<mark_weaver>I think we worked around the problem in guix by avoiding the affected procedures higher up in the stack.

<mark_weaver>it was in guix/serialization.scm, iirc

<civodul>i remember that one, not sure if it's related

<civodul>could be

<civodul>i'm trying with a newer Guile now

<mark_weaver>note that the bytevector fix is post-2.0.11

<mark_weaver>(although the debian package cherry-picked the fix)

<civodul>yeah i know

<civodul>for weeks i've been thinking: when do we release 2.0.12?

<civodul>i think it's in a state that we could release

<civodul>there are so many bug fixes in there

*sneek chants "Release! Release! Release!"

<mark_weaver>there are some remaining problems.

<mark_weaver>there was a change made in stable-2.0 that broke guile-gnome

<mark_weaver> http://bugs.gnu.org/18570

<civodul>it's g-wrap, so it must be a GOOPS-related change

<mark_weaver>also, I think that the new "foreign objects facility" API to replace SMOBs needs more discussion and review before releasing it.

<mark_weaver>it was committed into stable-2.0 less than 25 hours after the first proposal was posted

<mark_weaver>and I guess it's on me to continue that discussion now

<civodul>right, i had forgotten about that one

<dsmith-work>mark_weaver: What's wrong with SMOBs?

<dsmith-work>Other than C, of course.

<mark_weaver>dsmith-work: ask wingo, he's the one who decided to replace them.

<dsmith-work>Ah

<mark_weaver>dsmith-work: here's where he proposed them: https://lists.gnu.org/archive/html/guile-devel/2014-04/msg00069.html

IRC channel logs

2015-01-15.log