IRC channel logs

2014-05-08.log

back to list of logs

<amgarching>Given that files are most popular means of inter-process communication (in the broader sense) they are surprisingly inconvenient for anything real-time.
<amgarching>So to summarize, we have (1) (write data port) invoking (2) port-type specific write method cbop_write() which in turn invokes (3) user-supplied port-instance specific procedure write! (aka write_proc in C). Which one of these three should be responsible for buffering?
<amgarching>user-supplied is the only one that can be adjusted easisly, right.
<civodul>exactly
***sneek_ is now known as sneek
<nalaginrut>morning guilers~
<ksinkar>morning
<ArneBab>moin
<wingo>moin
<wingo>we have precise effects analysis for vectors now :)
<nalaginrut>cool
***vicenteH` is now known as vicenteH
<ArneBab>wingo: so vector-code is now better optimized?
<dsmith-work>Morning Greetings, Guileers
<amgarching>What could possibly go wron here http://pastebin.com/EE9xu2Y3 where two threads use two different ports concurrently? One is a file/fifo port another is "custom binary output port" from r6rs. I seem to recall some race/deadlock issue with port tables being discussed.
<amgarching>sorry if duplicate post, something seems to be wrong with my other IRC client.