IRC channel logs

2014-07-03.log

back to list of logs

<civodul>Hello Guilers!
<janneke>hello civodul!
<janneke>you asked me something yesterday and i simply asked guile
<janneke>(read (open-input-string "#{ foo \\"bar\\" #}"))
<janneke>$2 = " foo \\"bar\\" "
<janneke>...but i fail to `see' what you mean
<janneke>(i have parked the idea of (locals) and #{ ... #} for now, as i don't get it to work with lexicals)
<civodul>janneke: so #{...#} is essentially syntax for string literals?
<civodul>or does it allow escapes or something like that?
<civodul>i'm trying to understand what this is all about :-)
<amgarching>Hi, Guilers! What is supposed to happen when I "simultaneousely" start multiple interpreter processes with updated source that need to be compiled first? I am observing an unusually long delay with 8 processes compiling (they all do) the sources in comparison to 1.
<amgarching>Everything works, I just want to understand how. That is "mpirun -np 8 ./guile-script-with-custom-build-in-shebang"
<ArneBab>amgarching: they should firist run most scripts in interpreter-mode: It bootstraps itself
<ArneBab>amgarching: when you run it in parallel, many more files are run in much slower interpreter mode than when you run it serial
*ArneBab is just repeating stuff he read in here some time ago
<amgarching>My hypothesis is the processes compete for the some kind of lock on the *.go files and because of contention noone comes to rewriting them. I dont have number but the feeling is the startup time is more than 8 times up.
<amgarching>Also not sure if NFS is a part of the problem.
<amgarching>well there is no *problem*, just curiosity ...
<ArneBab>more than 8 times up is expected: if the interpreting-mode is 10x slower than the compiled-mode, and the *used* files aren’t compiled yet, then the parallel compile will be slower than the serial compile.