This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [tree-ssa]New fortran frontend


On Saturday 26 July 2003 11:33 am, Steven Bosscher wrote:
> Op za 26-07-2003, om 12:10 schreef Toon Moene:
> > Steven Bosscher wrote:
> > > All we use is GIMPLE, so if tree-ssa bootstraps on a target then so
> > > should gfortran.  That's one of the great advantages of a truely
> > > language/target independent IR :)
> >
> > Ha !  Wait until we implement ASSIGNED GOTO in the front end and watch
> > the house of cards come down :-) :-)
>
> *sigh* Old crappy fortran "features"...

Actually, by my reading of the standard it should be possibly to implement 
thease without resorting to the C "computed goto" extension. Shouldn't be 
too hard to transform them into select/goto. This avoids the problem of 
what happens if you assign to a kind=4 integer on a 64-bit machine.
Anyhow, this is getting OT.

> > BTW, Steven Kargl made a wise remark on the g95 development list: g95
> > needs the GNU MultiPrecision library (gmp).  I don't know if that's
> > correctly autoconf'd for, yet.
>
> It is not, because we haven't figured out how yet, and there is no
> standard test for it.

Last time we discussed this, I seem to remember thet a temporary solution 
was to just say "don't do this unless you have gmp installed". Obviously we 
need to find a proper solution though.

Paul


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]