IRC channel logs

2018-03-14.log

back to list of logs

<spk121>berndj: you could turn off buffering on the ports complete, so you don't have to flush them.
<berndj>spk121, yeah, but that seems even more icky :-/
<berndj>the problem goes away (apparently entirely) if i explicitly call scm_fluch_all_ports, i'm just surprised at the apparent necessity of doing so
<berndj>i find it really puzzling - i can think of no explanation more than "magic stuff happens". another atexit() handler still runs, so it isn't a total weird-out by the process
<berndj>is it possible that the program state during exit(3) is not compatible with the GC working correctly?
<spk121>as far as I know, there are many uncollected garbage objects at the end of execution. Guile doesn't promise to call the GC on everything at close.
<berndj>it's not that i want the GC to run, it's that the GC doing some stuff and the program flushing its output seem to be mutually exclusive
<berndj>both seem to happen inside an atexit handler
<spk121>oh, ok.
<berndj>and it did look to me like flush-all-ports would cons up some stuff, thereby triggering some GC stuff to happen
<daviid>I found and fixed the bug in gnu foliot, using guile-2.2, that was raising exceptions on certain entry callback
<amz3>héllo all
<amz3>spk121: interesting project, what's the plan?
<spk121>virtually all html links to SSAX tutorials are dead links
<spk121>amz3: it was an attempt at writing a simple 2D sprite game engine.
<spk121>the console is the debug console for the game