Re: [Cplusplus-devel]Re: macosx notes

[David Sugar]

I have done a slight re-write of BSD semaphore code (this is also used in 
linux targets) and perhaps I have made it a bit better since the destructor 
code will now only be used for semaphores that initalize successfully in the 
constructor.  It should hopefully now also do either sem_init or sem_open.  
If it works okay, we will need to merge this into HEAD as well.  We might be 
able to make it even better later by checking the errno from sem_init rather 
than depending on a preprocessor macro to support things like macosx which do  
not impliment sem_init.  In any case, it also needs lots of testing.

On Monday 31 March 2003 02:37 am, David Sugar wrote:
> I had worked on it a bit, but I still have not gotten it to run.  I tried a
> small change to use sem_open but I think the * returned by sem_open is
> meant to be used as if a rvalue, so conversion to _sempahore and
> referencing it by address does not directly work.  I have an idea to
> resolve this I will try shortly to support both sem_init and sem_open
> versions of sem_t.  At least the compile build issues should now be
> resolved so that it will be easier to get running on macosx.
>
> On Wednesday 26 March 2003 04:52 pm, Federico Montesino Pouzols wrote:
> > I have found in my mailbox of cc++ bugs still to be fixed that
> > sem_init does not work on macosx, which could explain the freezes. Ian
> > Gough, who has been testing cc++ on macosx may have some comments,
> > since I believe he was able to get cc++ working after some hacks...
> >
> >     An yes, I have not seen these kind of problems on FreeBSD.
> >
> > On Wed, Mar 26, 2003 at 11:21:22AM -0500, dyfet wrote:
> > > Curiously, with a relativily few simple changes I can get reconfig to
> > > work and (r1 based) common c++ to compile on macosx with a configure
> > > script and libtool built from macosx's distribution of libtool,
> > > autoconf, and automake.
> > >
> > > However, I have yet to get a demo or test to actually work on the test
> > > machine in question, at least from the build directory.  Anything that
> > > is linked through ccext2 (including the simple digest test app) appears
> > > to segfault in the dynamic linker while loading/resolving the image.
> > > Things that use just ccgnu2 just freeze...most curious, since the last
> > > time I had tested under FreeBSD (which is somewhat related), it
> > > generally worked very well there...
> >
> > -------------------------------------------------------
> > This SF.net email is sponsored by:
> > The Definitive IT and Networking Event. Be There!
> > NetWorld+Interop Las Vegas 2003 -- Register today!
> > http://ads.sourceforge.net/cgi-bin/redirect.pl?keyn0001en
> > _______________________________________________
> > Cplusplus-devel mailing list
> > address@hidden
> > https://lists.sourceforge.net/lists/listinfo/cplusplus-devel
>
> -------------------------------------------------------
> This SF.net email is sponsored by:
> The Definitive IT and Networking Event. Be There!
> NetWorld+Interop Las Vegas 2003 -- Register today!
> http://ads.sourceforge.net/cgi-bin/redirect.pl?keyn0001en
> _______________________________________________
> Cplusplus-devel mailing list
> address@hidden
> https://lists.sourceforge.net/lists/listinfo/cplusplus-devel