[sane-devel] Re: pthread_cancel and MacOS X

Jaeger, Gerhard gerhard@gjaeger.de
Sun, 11 Jan 2004 18:57:09 +0100

Hoh, hoh...

stop, I don't say, that it is not needed or not useful!!!
I only mean, that we should not spend too much time on a problem
which needs to be fixed outside of SANE. 
Also please remember:
Using the pthread approach instead of the Unix process stuff has also
been done, because of MacOSX is not able to inherit file descriptors
to a child process. Now that the sanei_thread stuff seems to work for
quite a lot of platforms, we ran into the next MacOSX trouble...
What comes next?
I mean I appreciate your work as I think it is useful to spread SANE
over another platform, but I don't think that SANE should work around
each OS bug, especially here, when there's no clean solution for this
problem.  This needs to be solved by the guys, that implemented the
pthread stuff there.
Henning, what do you think about changing the test-backend. As far
as I understood, the test-backend simulates, when the reader-process
has finished, a "hanging" backend, that is "killed" by the parent process.
Probably we can accept a workaround only for the MacOS platform...

Hope we find a good solution on that. I'll recheck some other approaches...

On Sonntag, 11. Januar 2004 15:28, Mattias Ellert wrote:
> Jaeger, Gerhard wrote:
> > Hi again,
> >
> > I also think, that a working test-backend is not that essential, as
> > working "real" backends.
> Here I must strongly disagree. A working test backend is extremely
> essential. It is a very important tool in debugging.
> If a frontend works with the test backend, but not with some other
> backend, you know that the problem is (most probably) in that backend
> and not in the frontend itself.
> If you can change the test backend in some other way so that the patch
> is not needed, that would be fine, but giving up a working test backend
> is not a viable solution to the problem.
> 	Mattias