IRC channel logs

2014-01-02.log

back to list of logs

<nalaginrut>morning guilers~
<adhoc>mornen
<nalaginrut>heya
<eg0>has a manager for site ever been proposed?
<eg0>it would be nice to have a simpler way to package up things
<eg0>and to install packages too, i guess
<mark_weaver>there's guildhall
<mark_weaver> https://github.com/ijp/guildhall/wiki/Getting-Started
<eg0>thar she blows!
<eg0>thanks!
<mark_weaver>np!
<nalaginrut>Is it a good idea to make Scheme as IR?
<XexonixXexillion>I'm trying to install guile reader, but it keeps throwing an error saying that it expects guile 1.8.x or higher (I'm running 2.0.9). It throws this error just after it fails to check for scm_take_u8vector. Any ideas?
<XexonixXexillion>It also gives a guile site directory that doesn't exist on my machine "/usr/share/guile/site/2.0" yet my guile is in "/usr/share/guile/2.0"
<jemarch>hi
<linas>Happy New Year, guilers!
<linas>Soo .. I need help.
<linas>I've been mangling a lot of UTF8 text (french, polish, lithuanian) in guile, and onece very blue moon, get an "encoding-error"
<linas>Can't figure out why. If I re-run the offending text, the error doesn't happen, suggesting that its some race condition.
<linas>My code is essentially single-threaded ... so this is mysterious.
<linas>How ight I begin to track this down?
<linas>(fwiw, its guile-2.0.9, hand-installed on a 5-year-old CentOS)
<mark_weaver>linas: why don't you start by using the guile debugger? get a backtrace from the encoding error and see what's going on.
<tupi>linas: do you (set-port-encoding! port "utf-8") on every port you use?
<tupi>do you use gnome or 'pure' guile?
<tupi>actually you need these 2: (set-port-encoding! port "utf-8") (set-locale port)
<tupi>if you use gnome, you must do it on (current-output-port) _before_ to load gnome modules
<mark_weaver>he's working on opencog. I doubt that gnome is involved here.
<tupi>so that setting gtkentries using like (format #f "~A" (_ "get my polish string")) would work
<tupi>have to go, bbl, good luck
<mark_weaver>there are a number of ways to set the encoding, the best of which is simply to set your environment variables for a UTF-8 locale and do (setlocale LC_ALL "") during your program initialization.
<mark_weaver>but the fact that the problem seems to be sporadic makes me think it's not something this simple.
<tupi>in this case he needs to catch a possible error [which happens if the user is not the admin of the computer, and sets his LANG to a non unstalled locale ... [beleive me it hapens :)]