IRC channel logs

2018-04-06.log

back to list of logs

<ArneBab_>manumanumanu: nice! SRFIs are great, and this could help them spread. Also it’s already multi-implementation
***dmiles is now known as dmiles[m
***dmiles[m is now known as dmiles
***fibratio` is now known as fibration
<amz3>Labu: hey
<amz3>Labu: what's up? Did yoiu go with sqlite or wiredtiger?
<lambda-lifter>I'm computing the root of a function (using newtow's method) and running up against
<lambda-lifter>guile's numeric tower. At a certain point, I get a flonum that is represented like this: 1.4287832697850099e-63
<lambda-lifter>If I compare this value agains the 1.00 flonum using (fl>? 1.00 1.4287832697850099e-63) I get
<lambda-lifter>#t which is incorrect. I also get the wrong resuts with other flonum procedures. What am I missing?
<OrangeShark>lambda-lifter: what version of guile?
<lambda-lifter>guile 2.0
<Efendi>Turk Yokmu Hec :D
<Labu>Hi amz3
<Labu>I realized that I didn't have the requird verison on my system
<Labu>so I returned on sqlite
<Labu>but I think try to compile guile 2.2 on my system later
<Labu>On debian (my another machine) guile is on 2.0.13 less than netbsd (2.0.14) version
<spk121>My Guile-based interactive fiction (aka choose your own adventure) game engine runs now. This has been a fun month's hack.
<spk121>Here's a video.
<spk121> https://github.com/spk121/burro/blob/master/game/screencast01.webm?raw=true
<OrangeShark>spk121: that is pretty cool :)
<rekado_>spk121: heh, I *am* impressed :)
<spk121>thanks!
<daviid>sneek: later tell cantonano please note that g-golf is experimemental still, has never ever been released ... use guile-gnome instead, which works perfectly still
<sneek>Got it.
<daviid>sneek later tell cantonano then please tell me and/or report the g-golf autotool related problem you claim you had - here or use the maling list - and claim you had to solve 'manually' (in your blog post), that doe not make sence to me
<sneek>Okay.
<daviid>sneek: later tell cantonano rt postgress, we guile-pg, very advanced binding but which does use the dynamic ffi, then guile-squee, which is also 'experimentl', checkout the devel branch, I autotool chained it ...
<sneek>Got it.
<daviid>sneek: later tell cantonano the autotool chain is a must, at the heart of the success of the entire GNU Porject
<sneek>Okay.
<daviid>sneek: later tell cantonano if you wan to check you have libpq, just do PKG_CHECK_MODULES([LIBPQ],[libpq >= 9.4.4]) - this is the guile-squee entry
<sneek>Will do.
<daviid>sneek: later tell cantonano OrangeShark wrote a template, which is awsome! here https://erikedrosa.com/2017/10/29/guile-projects-with-autotools.html and there are quite a few tutorials ... the thing is, one has to study a bit, and practice ... then ask here and on #autotools, they are very nice and very helpfull
<sneek>Okay.
<daviid>OrangeShark: as time allows, it would be nice to grow your awsome template with a tests (either guile-lib, which I prefer, or srfi-64 based) and a doc (texinfo based, I can help) subdirs and make the few appropriate changes to the existing template so it works (Makefile.am ...) - I know it's work, but for a beginner, having a working 'make check' example is extremely useful, as well as a doc template, with the verion automatically
<daviid>updated from the configure, a tiny pre structure and the doc license ...
<OrangeShark>daviid: I've been meaning to get on that, I was going to include it in the first blog post but it would have been too long. I already have tests using srfi-64 and texinfo docs implemented on my own project, but I wanted to do more research (like looking at guile-lib for tests).
<rekado_>I wonder what sxml-match brings us over plain match.
<rekado_>I found match to be much more flexible and usable than sxml-match.
<TMA>when compiling 2.2.3 --without-threads, I have run into reference to scm_pthread_mutex_lock remaining. here's patch: http://lpaste.net/4452868908189220864
<TMA>when compiling 2.2.3 on mingw64, there is a problem with missing <netinet/tcp.h> header. creating blank (or logically blank) file in guile-2.2.3/lib/netinet/tcp.h fixed that problem. however the approach used with netinet-in.in.h shall be probably used; my autotools magic is too weak to fix it, though
<TMA>the return value of start_child is incompatible in the .h (int) and .c (pid_t, presumably short); patch: http://lpaste.net/7868539382314041344
<TMA>when there is GUILE_AUTO_COMPILE=0 ../meta/build-env guild snarf-check-and-output-texi > guile-procedures.texi run during the build, it fails with:
<TMA>In procedure primitive-load-path: Unable to find file "C:/projects/foss/MINGW-packages/mingw-w64-guile/src/build-x86_64-w64-mingw32/libguile/C:/projects/foss/MINGW-packages/mingw-w64-guile/src/build-x86_64-w64-mingw32/meta/guild" in load path
<TMA>it looks like it pastes the full, absolute, file name to the search path. however the is_absolute_file_name in load.c seems to detect C:/... as absolute, I did not find why it is appended to the search path
<TMA>if anyone of you think these reports are worth anything, post them to the bug tracking system. Patches are trivial enough that there should not be any copyright concern, in any case I release them to CC0/public domain.
<daviid>sneek: later tell cantonano wrt g-golf, here even 'make distcheck' pass
<sneek>Okay.