Re: gfortran: OK, I quit, you win
From: Gordon Sande (g.sande_at_worldnet.att.net)
Date: Mon, 17 Jan 2005 20:27:08 GMT
> James Van Buskirk wrote:
>>I wanted to report a [gfortran] bug through channels, but the
> channels were
>>even worse than Intel's; I didn't think that was possible!
> You can send an email to fortran at gcc dot gnu dot org , which will
> soon be read by gfortran developers. The mailing list archive for the
> current month is at http://gcc.gnu.org/ml/fortran/2005-01/ . I am sure
> they are not trying to defeat you :) . It is your demanding codes that
> often defeat the compiler writers!
When I found a bug in gfortran I tried the official reporting scheme.
BUGZILLA!! (I stopped admiring the kiddy cartoons long ago)
I was informed that I needed an account with a password. Since I
already have a great excess of such I declined and chose to use the
bug reporting script that was offered as an alternate but it was a
broken link. So I sent an email and was told that the gfortran
supplied through fink was so out of date that my bug had long since
The second time around I was wiser so I just posted the bug to c.l.f
and one of the developers thought it was a neat problem. It echoed
for several weeks (the select case type problem).
It turns out that my, or JVB's, low opinion of their bug reporting is
polite compared to what the developer said about it.
On the one hand it a nice public service that is being provided
but on the other hand (sci.econ anyone?) they do seem to make it
hard for anyone other that the most committed to either access their
product or report usage experience with it. When it is pre-alpha that
is OK but when it is coming up on public release it ceases to be OK.
Somewhere along the line a bit of effort to smooth things out would
pay great dividends. You are hearing repeated suggestions that such
a time has now been passed. The complaint is really a recognition of
a success so the they need not be offended but rather should be pleased
that they have gotten to where such things become an issue.