On Monday 13 October 2014 02:54:37 Jon Simons wrote:
> On 10/13/14, 2:38 AM, Jon Simons wrote:
> > On 10/12/14, 10:39 PM, Andreas Schneider wrote:
> >>>> I'm happy to rework any of this as needed, any feedback is very
> >>>> welcome.
> >>
> >> For feedback, take a look at http://test.libssh.org/
> >>
> >> Hint: It is red ;)
> >
> > Ah, my bad -- it looks like I've broken the build for a few targets that
> > I've never built before.
> >
> > I'm taking a look right now to see if I can fix these quickly, hopefully
> > I will have a patch with some progress to send out shortly.
>
> Sorry for the second mail, I don't mean to spam -- I am fixing the build
> breakages which are due to pkd. There are a few cases where the build has
> -DWITH_SERVER=OFF set, in which case we do not want to build the pkd test,
> because it relies on all the server bits. And I have some porting to do
> for the FreeBSD build, too.
>
> However I believe that the actual test failures are due to this new change:
>
>
> http://git.libssh.org/projects/libssh.git/commit/?id=250f50648755415a138b13
> 47a17a367ad384b66d
>
> I think that change is not correct: for a non-blocking connect(2) it could
> be that the initial call returns -1 along with errno set to EINPROGRESS.
> I suspect that is what is happening, but with that diff, that case will be
> considered an error. I have not looked deeply at it yet, though.
Yes, I guess you're correct. The patch is incomplete and needs additional
checks on the errno.
I try to find some time during the conference today to fix it. If not maybe
later this week.
Thanks for pointing it out!
-- andreas
