This is the mail archive of the gcc@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: GCC 4.0 Status Report (2004-01-05)


 I must voice my opinion here, some 'old' g77 fortran code is really braindead as for
 using large common blocks and similar instead of the mechanisms that are in place, also
 stuff that's deprecated like hollerith constants and similar is of historical interest
 only.

 Thus I do believe that as long as the compiler adheres quite closely to the standard and
 does produce decent code the complaint 'it doesn't compile my old code' is
 in many cases rather worthless if not accompanied with an apropriate analysis of
 where the problem really is.

 However, it's a bit sad to hear something like this since the gfortran team has been
 making a major effort in order to get gfortran to compile g77 code, i think the people
 involved has made a tremendous job getting gfortran good enough for 4.0 .

 So if there is some things that are miscompiled bugreports or code examples or something
 similar is constructive critisism, while 'gfortran doesn't compile my old shitty
 program' because of it's antics is just pulled out of the air.

 I really apologize if this comes off a bit to rough, but this is a technical
 discussion list and unless there is solid data to back things up it's just a major
 put down to read stuff like this.

 I have ported some portions of old code to gfortran and it has so far never been too
 hard, ( one exeption or two, but I do realize that that code is old and rotten and
 really merits a rewrite ).

 So finally, I don't really think this is an issue, if you look at the progress that
 have been done on the libs, and the test suites ( LAPACK, BLAS and such ), in the
 last weeks it amounts to some major improvements. Also for some code I have a reduced
 running time, ( executes in 30% of the time of the g77 compiled binary ), which for
 most applications where fortran is used is worth more than anything.

 I think that as long as the fortran 77 and fortran 95 standards are closely respected
 'legacy' compability is not all that important.

 All this is only opinions, but what I wanted to state was that gfortran is very good
 in it's current incarnation, and getting better all the time as for g77 compability
 so I can't really find that much merit in the complaint, give some specifics of the
 shortcomings stated please.

 / regards, Lars Segerlund.

On Mon, 10 Jan 2005 01:24:28 +0100
"Giovanni Bajo" <giovannibajo@libero.it> wrote:

> Geert Bosch <bosch@adacore.com> wrote:
> 
> >> I think that at least known wrong-code bugs for primary release
> >> platforms should be fixed or documented as such.
> >
> > There is nothing special about known wrong-code bugs versus
> > unknown ones. For a new technology, like gfortran, users
> > will find lots of bugs after the first release. There will
> > be tons of known wrong-code bugs.
> 
> This is where our disagreement is. You can rewrite as much as you like of GCC,
> but if there is something which makes the user unable to compile its code with
> the new version, while it works well in the old version, then it is a
> regression. As a user, I don't care if you rewrite everything or just keep
> fixing bugs.
> 
> Giovanni Bajo
> 


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