Re: shm status

[Christopher Faylor <cgf at redhat dot com>]

On Sun, Jun 09, 2002 at 10:18:20PM -0400, Charles Wilson wrote:
>Except that I think Linus hates the /proc interface -- all that text 
>parsing to pass info back and forth.  Binary data should be binary...

Actually, I think it's quite the contrary.  I've never seen Linus express
any dissatisfaction with all of that text, although I have seen plenty of
grumbling about it in the linux-kernel mailing list.  From time to time
someone proposes a /binproc (or whatever) interface.  I don't recall Linus
ever encouraging this, though.

FWIW, I don't like the text myself.  I just like going with established
interfaces in cygwin.  Reinventing the wheel is fine for other products
but it's not great for a product which is supposed to be emulating
the way things are done in unix/linux.

>Anyway, concerning ipcs and ipcrm...there are implementations of them in 
>cygipc -- but *do not copy them*.  It's that whole copyright thing.
>
>ditto the utilities in cygutils: msgtool.c  semstat.c  semtool.c  shmtool.c.

Actually, we have plenty of GPLed tools in the distribution, so there is
no reason not to use them.

They can't be part of the base cygwin distribution (i.e., they can't live
in winsup/utils) but that is no hardship.

I can't think of any reason not to take a look at the linux version of
ipcs and see how it makes things happen, either.  FWIW, when I do an
strace (on linux) on ipcs -a, it seems to just be doing a bunch of
shmctl's.  I don't think a /proc interface is really needed.

>However, that doesn't stop you from compiling them, linking them against 
>cygserver, and using them to help test and develop cygserver...

In reality, there is no reason why we couldn't include cygipc in the cygwin
release.  I was just concerned that the inclusion of cygipc would hinder the
development of a true cygwin DLL solution that used some of the principles
embodied in Robert/Egor's cygserver.

cgf