IRC channel logs
2017-05-12.log
back to list of logs
<daviid>ACTION finally got image histograms working in guile-cv, it needs fine tuning though: (a) too slow), (b) some /tmp latex names clashes may still occur... but getting close <daviid>ok, im-histogram is bug free now, here at least :), good. it is a bit slow, but because of it's calling pdflatex for some of the legends stuff, not because of guile, and i have some ideas to make it faster anyway <manumanumanu>nalaginrut: no. delimited continuations are better in every way <manumanumanu>ArneBab_: neat :) Next up: python2. We need to cut runtime by at least 60% O_o <manumanumanu>ArneBab_: your simple version is not the "fast" simple version, nor the correct one. the string every checks for #\\space \\t and \\n, but what we should just check is string-null? since read-delimited won't return any string with #\\space #\\newline or #\\tab. <ArneBab_>damn, then I copied the wrong version there. Can you pastebin the simple version (I’ll update it this evening)? <manumanumanu>ArneBab_: I just found out I can fake cond-expand in chez and make Nietzsche fast in both guile and chez without much fuss. Chez looks for modules with the extension .chezscheme.sls before anything else :) <manumanumanu>so, finally I can avoid using full continuations in guile for my coroutine generator <wingo>ACTION highfives manumanumanu re: "delimited continuations are better in every way" <manumanumanu>wingo: Well, they are! I used to thing Oleg was just a grumpy old man until I saw the light. <manumanumanu>wingo: I am trying to crowbar fibers into my next project btw. I am pretty sure I don't really _need_ to, but what's the fun in that? :) <wingo>please do fork the project and figure out things that could be better! <sneek>Welcome back civodul, you have 3 messages. <sneek>civodul, bavier says: updating OpenMPI is on my todo list <sneek>civodul, efraim says: 9 hours for make with guile2.2 on aarch64 <wingo>lol sorry for causing the pounce :) <wingo>i guess that's a guixpouce tho <efraim>i was wondering why i got tagged in #guile ;p <wingo>efraim: now that you are here :) that guile 2.2 on aarch64 was without prebuilt/, is that right? <efraim>~6 hours for guile 2.2 without prebuilt, 9 hours for guix with guile2.2 <efraim>I think it was 45 minutes for guile2.2 with prebuilt <ArneBab_>is there a guile-2.2 package for ubuntu (in some repo?) <manumanumanu>Hey... Why doesn't the make-coroutine-generator from srfi-121 work in guile? <manumanumanu>That works in both chicken (with small changes) and chez <manumanumanu>I have a slightly different version that didn't work and I thought I had ust done something stupid. <manumanumanu>or rather: it continues to yield the last result and never leaves <ArneBab_>manumanumanu: did the first auto-compile the code? <ArneBab_>did it show the ;;; <I am compiling stuff> message? <ArneBab_>I noticed runtime differences when it does that <amz3`>I want so much +to copy datomic <amz3`>nalaginrut: congrats for the release <amz3`>nalaginrut: did you play with datomic? <amz3`>nalaginrut: I read it's the hottest database un clojure land <nalaginrut>amz3`: I'm thinking about more cool new features <nalaginrut>amz3`: oh, datalog! which was what I want to add to Artanis <nikita1>nalaginrut: pdf doc is also needed regeneration without broken org-mode. <nalaginrut>amz3`: I'm considering the proper way to go. Write a datalog like dialect in s-expr. or just implement datalog as a language frontend <amz3`>nalaginrut: that would be nice <amz3`>nalaginrut: guile-log implements a datalog <amz3`>but you can "only" dump a given database to scm and reload it to memory, there is no notion of ACID <amz3`>there is no ACD guarantees, prolly it works in certains circustences <nalaginrut>amz3`: dump to scm and reload seems not efficient <nikita1>may be it's a common problem for pdf documentation that code samples (monotyped) are almost always rut out of page boundaries. <nikita1>I saw it in many GNU docs generated automatically .. <amz3`>nalaginrut: that's what I thinkn too <ArneBab_>nikita1: I typically need to adjust the code to be nice in PDFs — and reduce the font size a bit. <ArneBab_>being serious about a 72 char width limit helps, though. <nalaginrut>nikita1: all the docs are generated automatically by org file, I have to do so, since GNU standard requires that bunch of files <nalaginrut>the colored html doc is the easiest to keep good quality <nalaginrut>other files are generated by texi which is generated by org <nalaginrut>nikita1: but I'm still willing to fix the pdf if you could point me out the problem <paroneayea>I'd love to see more skribilo support, especially nicer info output from skribilo <paroneayea>I'd love to use skribilo for the default for a few of my projects... I'd prefer it over writing texinfo. <paroneayea>the info output in skribilo is not nearly as good as texinfo's though, at present. <daviid>for info, texinfo is mandatory for GNU projects, and texinfo is a must, also we have stexi ... which is awesome <paroneayea>daviid: I know this because mediagoblin was allowed to use sphinx as long as we had an info output. <paroneayea>similarly, autotools is not required, but the interface that autotools currently provides (./configure && make with associated flags) is <daviid>paroneayea: ah ok, texinfo is better imo anyway, but ... as well as the autool chain, quite a lot better actually, but i guess all this is taste (and knowledge experience) <daviid>the only missing bit in texinfo is proper support for goops <daviid>i mean proper support for doc about generic function multimethod polimorphy dispatch systems...