Re: gsl-0.41 and SparcWorks C

[Bernd Petrovitsch <bernd@ict.tuwien.ac.at>]

In message <76vhaynne7.fsf@odie.lanl.gov>, Mark Galassi wrote:
>This was fixed recently, and I wonder if you could tell me if the
>version in anonymous CVS works for your compiler.

Uuhh no, apparently the "-Wp,-MD,.deps/$(*F).P" gives at best 
strange error messages with the SparcWorksCompiler - seems to be
a gcc-feature.


In message <76hfmim0f4.fsf@odie.lanl.gov>, Mark Galassi wrote:
>I can imagine that being true, but I wonder if it's still true with
>the new optimizations that have come in with egcs.  I'm sure the egcs
>lists have discussion of that :-)

My experiences (i.e. performance test) are with gcc-2.7.2 (and I suppose
that 2.8 is not much better) and a CPU-only program (almost no memory or
disk acceses, only number crunching). In the recent iX (a monthly german 
computer journal http://www.heise.de/ix/) they write that the maintenance
of the "official" gcc was handed over to the egcs people since it was
pretty silent around gcc the last years. First version should be
named gcc-2.95 (which relates to the planned egcs-1.2) and should
be released on July 8 (they write in that form there, so if someone
knows better, please write).
At them moment I have here gcc-2.8.1 installed, so which program(s)
should be ran and measured ?

>    Bernd> "end-of-loop code not reached" on every GSL_ERROR
>    Bernd> statement. Apparently it detects the "return" in the
>    Bernd> do-while loop and does not check that while(0) does not
>    Bernd> yield any code (hopefully :-).
>
>That's a pity that you get a warning: the do { ... } while (0); is a
>trick used in assertion macros in several major pieces of software.

Yup. I know. Actually I don't see a reason for not just using
{ ... }  because it works here too (as these issues and discussions
are too academic IMHO).

	Bernd

PS : The SparcWorksCompiler has a "-v" option (for stricter semantic
checking) and then it complains about e.g. empty statements.

--
 Bernd Petrovitsch               Institute of Computer Technology
 Gußhausstraße 25-29, A-1040 Vienna Email: bernd@ict.tuwien.ac.at
 "Der Horizont vieler Menschen ist ein Kreis mit Radius Null -
  und das nennen sie ihren Standpunkt."               A. Einstein

PGP signature