Re: gfortran: OK, I quit, you win
From: James Van Buskirk (not_valid_at_comcast.net)
Date: 01/18/05
- Next message: FX: "Re: gfortran: OK, I quit, you win"
- Previous message: Jan Vorbrüggen: "Re: closures"
- In reply to: FX: "Re: gfortran: OK, I quit, you win"
- Next in thread: FX: "Re: gfortran: OK, I quit, you win"
- Reply: FX: "Re: gfortran: OK, I quit, you win"
- Reply: FX: "gfortran, logical(8) (was: gfortran: OK, I quit, you win)"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Tue, 18 Jan 2005 08:07:25 -0700
"FX" <coudert@alussinan.org> wrote in message
news:csim1f$2001$1@nef.ens.fr...
> > Woah thar, hoss! What is a MinGW tree?
> So, let's say we want to be constructive. I replaced "tree" by
> "directory", and added information about how to install MinGW.
Good. I hope that my having related my experiences regarding
installing MinGW helped inform you about the utility of this
step.
> > How do I know that I have one?
> Cause you know what's on your computer?
I Windoze-land nobody has a clue about what's on their computer.
Look at the statistics about how many Windoze computers have
adware/spyware installed on them. Even for a user who cares
about what's on his PC, there's absolutely no way to keep
track of what everything does, what everything is, what is
just junk that should have been deleted when some long-forgotten
package was uninstalled. Maybe in UNIX-land there is some
way to audit everything so you can figure out what doesn't
belong anymore, but you really get a feeling of helplessness
when contemplating doing this on a Windoze machine.
> > it might be the mythical MinGW tree.
> Could you please refrain from being (to much) sarcastic? People are
> working to get this result, and other people are finding it useful. Many
> have constructive criticisms to make, and they are taken into account. If
> you feel like using gfortan right now is too difficult, then wait for a
> stable release.
I recall a cartoon where the waiter is presenting a customer
with his choice of dueling pistols and the customer's date is
saying "I told you that you shouldn't have tried to order in
French." It seems that the general tone of my communication
doesn't translate all that well out of the 'Murkin language,
but I am trying to help as well. What really got me interested
in this exercise was a discussion in this forum about
http://home.comcast.net/~kmbtib/Fortran_stuff/logical_gen.f90
and the file it creates, logical_all.f90. Someone showed me
the output he got with gfortran and I could see that it was
incorrect. He said that he couldn't cut the program down to
reasonable size while still maintaining the incorrect behavior,
and suggested that I try doing so, but we were at that point
at an impasse because I think he, and most if not all of the
gfortran developers didn't see the point of my example, and
I couldn't install gfortran on my machine. Now, thanks to
your efforts, I could install gfortran, and from the exertions
required on your part you can likely see that it would have
been quite a long shot for me to have attempted the to do so
on my own. I have been able to show that the bug reproduces
in a smaller example, so I would like to see what this third
party actually attempted; it seems that most posters other
than myself consider it bad form to post complete compilable
examples. But I think this paragraph is not news to you as
I seem to recall your participation in that thread.
> > When I did this it ended up putting everything in my C:\gfortran_stuff
> > directory instead of creating a C:\gfortran_stuff\gfortran-mingw
> > directory like it did the first time.
> Yep. This shouldn't change any more.
This didn't translate well into English. I have no idea what
this means: will gfortran no longer attempt to create its own
directory tree, or does the directory tree not get rebuilt
after the first installation, or something else?
> > gfortran now worked!! (sort of)
> Yes, sort of. gfortran is still alpha software.
And this makes me uncertain about the goals of gfortran.
Certainly it will require titanic effort on the part of one
or more individuals to make it fly, and the effort, considered
as a training exercise, might be the ding an sicht here, but
if it's trying to find some kind of niche, what is it? A
free Linux compiler is already available for noncommercial
use by Intel, and it's going to be tough to beat, especially
in terms of optimization. It would be way cool if it were
usable on platforms the commercial vendors or even g95 won't
touch, but it isn't targeted on 3 out of the 4 platforms I
can boot to at home (QNX -- yeah, Fortran on your cell phone!,
OS/2 -- even ATMs need Fortran, Alpha-NT -- I have CVF 6.5
for that in any case) some of these embedded environments
would be quite tempting targets to me if I were a gfortran
developer (which I could never be: no GPL for me!) since
platforms tend to grow from the small towards the big, from
embedded to mainframe, rather than the other way around, if
you get the vision thing. g77, for example, never ported
to QNX, and I don't think Watcom ported their f77 compiler,
either, just their C/C++ compiler.
> > http://home.comcast.net/~kmbtib/gfortran_bugs/
> I won't look into it regularly. Using the fortran@gcc.gnu.org
> mailing-list is one sure way to report bugs (see below), if you don't
> want to use bugzilla.
No need to poll it. If I have some stuff that I consider really
important I'll post a notice here.
> > Does the mailing list mung your email address for you by default? I
> > can't put my unencrypted email address in a public forum and it's a
> > royal PITB if I have to do it by hand every time.
> I use my usual mail account for this mailing-list (unlike the
> comp.lang.fortran newsgroup), and it is transformed into "coudert at lcp
> dot u-psud dot fr" as usual. I don't get any spam (and I've been there
> for a few months now), I even don't need to run spam-detection software.
Well, the creative spammer could still harvest that, and even
gfortran can't decipher my sig.f90 file!
-- write(*,*) transfer((/17.392111325966148d0,6.5794487871554595D-85, & 6.0134700243160014d-154/),(/'x'/)); end
- Next message: FX: "Re: gfortran: OK, I quit, you win"
- Previous message: Jan Vorbrüggen: "Re: closures"
- In reply to: FX: "Re: gfortran: OK, I quit, you win"
- Next in thread: FX: "Re: gfortran: OK, I quit, you win"
- Reply: FX: "Re: gfortran: OK, I quit, you win"
- Reply: FX: "gfortran, logical(8) (was: gfortran: OK, I quit, you win)"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
|
