IRC channel logs


back to list of logs

<manumanumanu>civodul: haha! I am happy I was right, but I am sorry to hear that you have to debug it :(
<manumanumanu>I have only met it on the implementation side of things, and I can imagine weak references suck even more to have to debug from the other side
<civodul>indeed :-)
<manumanumanu>Hmmm. regarding fluids. Can I set a fluid in a module and have it active in code that imports that module? I am trying to change current-reader...
<civodul>yes, though if modules are compiled, they're obviously not going to be 'read'
<civodul>so you might want to use it in 'eval-when'
<dsmith-work>Hey Hi Howdy, Guilers
<manumanumanu>Why isn't read-hash-extend generalised? It would be neat to be able to hijack the reader for any kind of chars. Or maybe I'm just bitter because I can't install guile-reader
<manumanumanu>I give up. hijacking the current reader is no fun.
<dsmith-work>manumanumanu: Mostly history I guess. Guile's reader was (is?) mostly in C (as was most all of Guile).
<manumanumanu>dsmith-work: so I have noticed
<dsmith-work>manumanumanu: Take heart! As the compiler gets better and better, more and more of Guile is being rewritten in Scheme.
<manumanumanu>dsmith-work: I already started writing my own
<manumanumanu>:D :D
<ArneBab_>manumanumanu: I now ran our wordcount tool on the docker provided by , but performance is pretty bad (almost last place).
<ArneBab_>manumanumanu: how to repeat that:
<ArneBab_>manumanumanu: well … on the second run performance is about as good as nodejs
<ArneBab_>wingo: I see a factor 2 disparity between the first run of a script (where it gets auto-compiled) and the second run. Is that a known problem?
<ArneBab_>wingo: this is with 2.2.2
<ArneBab_>manumanumanu: about as good as nodejs *typescript*, native JS is faster
<manumanumanu>ArneBab_: how big disparity?
<manumanumanu>The compile shouldn't be more than 0.5s
<manumanumanu>ArneBab_: that v8 ds faster shouldn't come as a surprise
<manumanumanu>v8 is _blazing_ for single thread stuff
<ArneBab_>manumanumanu: I get roughly twice the runtime in the process which does the auto-compile
<ArneBab_>first run takes about 2x as long as second run (or subsequent runs)
<ArneBab_>manumanumanu: not in this test … Python2 is around factor 3 faster