Samba build failures on MacOS X

[Volker Lendecke]

On Mon, Jun 30, 2014 at 02:58:25PM +1200, Andrew Bartlett wrote:
> > IIRC the fork issue is not just weird, it breaks some fundamental
> > assumptions done by samba. You can probably get a toy build working, but
> > that's all.

Very likely smbd will just run fine and the OS/X documentation is just
laziness on Apple's part. I doubt that they deliberately break core APIs
that have been around forever, but it's highly likely that the rich GUI
APIs really break after a fork due to missing correct pthread_atfork
calls. Making a library that uses threads fork-safe is really, really
hard.

The problem is that a single commit can make a core API like getaddrinfo()
behind the scenes use a rich network API that never was supposed to work
across fork. At that point we're screwed without knowing. Apple won't
even listen to a bug report about that, pointing at the documentation.

> > I am conflicted on whether we should accept patches to make it build
> > knowing it is fundamentally broken (for the file server at leats), or if
> > we should get it to build and then take blame because it doesn't work
> > well.
> 
> Yes, I agree this is difficult, and the official man pages makes for
> depressing reading.  Yet the workaround those determined to use MacOS
> use is to build 3.6, which seems to me to be liable for just as much
> trouble. 

If we had a true maintainer of this with clean patches, I'd be more than
happy to bring them in.

Volker

-- 
SerNet GmbH, Bahnhofsallee 1b, 37081 Göttingen
phone: +49-551-370000-0, fax: +49-551-370000-9
AG Göttingen, HRB 2816, GF: Dr. Johannes Loxen
http://www.sernet.de, mailto:kontakt at sernet.de