gfortran

[Steven Bosscher]

On Thursday 21 December 2006 15:51, Eef van Beveren wrote:
>  L.S.
>  Fedora 6 has been installed on my computer
> and with it gFortran. It is shocking to find
> all my routines, build for g77, no longer
> running. I tried to repair some. But, I found
> that it is a hopeless task. There are even
> inconsistencies with regard to backslash
> (used to be always 2 backslashes when 1 is
> needed, but now it seems to depend on how it
> called. Hence, sometimes 1 sometimes 2 and
> God knows when 1 and when 2!). Thousands of
> lines of Fortran routines nust again be
> debugged. Beyond possibility, I do not have
> time for compiler builders who are joking with
> me. Please make g77 again available!!!!!

This is outright insulting.  Here, you have a group of volunteers
trying to basically "rescue" Fortran in GCC, and what do you do?
You assume they are "joking" with you.

Perhaps you're unaware that G77 was old, in maintenance mode, and
effectively abandoned.  The number of users GNU Fortran is pretty
high, but the number of people *helping* to keep GNU Fortran alive
is not in proportion, especially not compared to some of the other
languages supported by the GNU Compiler Collection.  For GCC 4.0,
Fortran would probably not even have been a supported language,
if this handful of people had not helped out developing gfortran. 
Tens of thousands of lines of Fortran compiler had to be written
from scratch to get gfortran to the point where it is now.  This
effort dwarfs your problems, I'm sorry to say.

(See also the SLOCcount results at the bottom of this message for
an estimate of the economic value of this effort, excluding the
test suite).

The scientific community is a lousy user base to develop compilers
for, if you are a developer for fun, as a hobby, like pretty much
all gfortran developers (including myself until recently).  Many
people in the scientific community whine, complain, and insult.
Almost none of them ever funds or contributes to gortran to help
improve it, or at least to fix those inevitable bugs in an entirely
new compiler that affect their code.

Apparently you are one of those thousands of anonymous GNU Fortran
users who take a free Fortran compiler for granted, *demand* one
even to be available.

Think again.

I like to compare this attitude to that of people holding up their
hand like beggars, expecting their hand to be filled every day, and
then feeling insulted when some day they find their hand empty at
the end of the day.

I understand that it is inconvenient for you that gfortran does not
compile your code at this time.  I think everyone is sorry that it
is impossible to write a compiler from scratch without having some
backward compatibility issues.  I also think everyone is hopeful
that when such issues are reported, they can be resolved quickly so
that all functionality of G77 is retained in gfortran.

You are fortunate that this group of talented and helpful people on
the mailing list will try to help you in their own free time.  You
could show a little more gratitude for that.

I hope your code will compile without problems soon.

Gr.
Steven





SLOC    Directory       SLOC-by-Language (Sorted)
68861   fortran         ansic=68861
66321   libgfortran     ansic=64324,f90=1891,sh=106


Totals grouped by language (dominant language first):
ansic:       133185 (98.52%)
f90:           1891 (1.40%)
sh:             106 (0.08%)




Total Physical Source Lines of Code (SLOC)                = 135,182
Development Effort Estimate, Person-Years (Person-Months) = 34.55 (414.64)
 (Basic COCOMO model, Person-Months = 2.4 * (KSLOC**1.05))
Schedule Estimate, Years (Months)                         = 2.06 (24.70)
 (Basic COCOMO model, Months = 2.5 * (person-months**0.38))
Estimated Average Number of Developers (Effort/Schedule)  = 16.79
Total Estimated Cost to Develop                           = $ 4,667,738
 (average salary = $56,286/year, overhead = 2.40).


generated using David A. Wheeler's 'SLOCCount'.