From: Dirk Koopman (terhi.victor@logonet.com)
Date: Mon Jan 03 2000 - 13:30:05 EET
On a heartfelt personal note, can I say that IPC message channels are one of
the hardest (read nastiest) programming things to get absolutely correct in
all circumstances.
I once wrote a very nice commercial spooler using them and although it works
very well, under the miriad fault conditions with client programs dying or
programmers using the api not closing their end properly etc, the small
buffer size and impossibility of knowing whether a asyncronous process (not
in your process group) has gone away without checking all the time means that
this sort of problem happens a lot.
Some time this year it will be converted to sockets and select.
On the other hand using a small 4 byte IPC shared memory segment provided the
easiest way to keep an absolutely reliable handle on how many people are using
something concurrently... (Hint: don't use the memory area for anything at
all...)
On 03-Jan-2000 Tomi Manninen OH2BNS wrote:
> I seem to have vague recollection that clx has had problems with IPC
> channels as well. I'm not sure however as I don't run it my self.
>
Dirk G1TLH
-- Dirk-Jan Koopman, Tobit Computer Co Ltd At the source of every error which is blamed on the computer you will find at least two human errors, including the error of blaming it on the computer.
This archive was generated by hypermail 2b29 : Mon Jan 03 2000 - 13:30:38 EET