Mail Archives: djgpp-workers/1999/03/15/20:45:31
Bonjour
> Alain Magloire wrote:
> > I tend to agree with you to keep most of libsocket separated
> > but it would probably make it easier for Richard D. to have some
> > hooks in the C lib.
>
> To be honest, I don't mind too much. libsocket comes with a libc patch that
> adds the error codes. Given the fact that libsocket doesn't work under DOS, it
> is probably best that libsocket isn't integrated into DJGPP.
>
Ack.
> After I asked what people thought about integrating libsocket into DJGPP, I
> got the impression that it would be OK. Maybe I'm just confused.
>
My impression was that DJ was "open" to the idea, if you could provide
libsocket as an addon, since libsocket was still not fully mature
and could not work *as is* on "vanilla DOS".
Of course I maybe completly of track and missed some emails.
> BTW are error codes like ENOBUFS really networking-only errors? I've found
> several occasions on which I wanted to use this error in my code in a
> non-networking context. Is this wrong?
In my P1003.1g draft :
ENOBUFS : Insufficient resources were available in the system to perform
the operation.
Even if the def. is general, it is networking specific.
I think, it originates from the BSD networking stack were MBUF
(Memory BUFfers) structures were use as some sort of message passing
(not in the IPC sense) between different layers. When the networking
internal code could not get no more memory it returned ENOBUFS.
--
au revoir, alain
----
Aussi haut que l'on soit assis, on est toujours assis que sur son cul !!!
- Raw text -