I have a question regarding the switch from gcc3.x to gcc4.x in OpenSuSE 10B3. I recently made this switch to fix a problem with my X server, but had to revert back to 9.3 because there is no g77 for 10 as gcc4.x uses gFortran. This is going to cause major problems for people in statistics and engineering as we often use a lot (and I mean a LOT) of legacy code written in F77. For instance, try removing g77 and then installing R-2.1.1 (http://cran.r-project.org) from source. It will not work, because there is a lot of legacy fortran code used. Is there a plan for this transition, or are we going to have to spend lots of time and effort recoding the legacy code into a more modern language? Thanks!! Dave H -- David A. Henderson, Ph.D. 1215 Dexter Avenue North, #535 Seattle, WA 98109 206-794-8552 DNADave@ComCast.Net
David Henderson wrote:
I have a question regarding the switch from gcc3.x to gcc4.x in OpenSuSE 10B3. I recently made this switch to fix a problem with my X server, but had to revert back to 9.3 because there is no g77 for 10 as gcc4.x uses gFortran. This is going to cause major problems for people in statistics and engineering as we often use a lot (and I mean a LOT) of legacy code written in F77. For instance, try removing g77 and then installing R-2.1.1 (http://cran.r-project.org) from source. It will not work, because there is a lot of legacy fortran code used.
Is there a plan for this transition, or are we going to have to spend lots of time and effort recoding the legacy code into a more modern language?
Thanks!!
Dave H should send this on the opensuse mailing list :-)
jdd -- pour m'écrire, aller sur: http://www.dodin.net http://valerie.dodin.net http://arvamip.free.fr
jdd sur free wrote:
David Henderson wrote:
I have a question regarding the switch from gcc3.x to gcc4.x in OpenSuSE 10B3. I recently made this switch to fix a problem with my X server, but had to revert back to 9.3 because there is no g77 for 10 as gcc4.x uses gFortran. This is going to cause major problems for people in statistics and engineering as we often use a lot (and I mean a LOT) of legacy code written in F77. For instance, try removing g77 and then installing R-2.1.1 (http://cran.r-project.org) from source. It will not work, because there is a lot of legacy fortran code used.
Is there a plan for this transition, or are we going to have to spend lots of time and effort recoding the legacy code into a more modern language?
Thanks!!
Dave H
should send this on the opensuse mailing list :-)
jdd
If the only change is gFortran and g77 could you create a link "ln -s gFortran g77". Would this allow his programs to compile correctly. -- 73 de Donn Washburn Hpage: " http://www.hal-pc.org/~n5xwb " Ham Callsign N5XWB Email : " n5xwb@hal-pc.org " 307 Savoy St. HAMs: " n5xwb@arrl.net " Sugar Land, TX 77478 BMW MOA #: 4146 - Ambassador LL# 1.281.242.3256 " http://counter.li.org " #279316
Donn:
If the only change is gFortran and g77 could you create a link "ln -s gFortran g77". Would this allow his programs to compile correctly.
The problem with this approach is that from all of the documentation I could find, gFortran is only a F90 and F95 compiler. So, is the gFortran compiler going to still compile F77 code, under the approach you describe above, or is the problem going to be much more difficult... Thanks!! Dave H -- David A. Henderson, Ph.D. 1215 Dexter Avenue North, #535 Seattle, WA 98109 206-794-8552 DNADave@ComCast.Net
David Henderson wrote:
Donn:
If the only change is gFortran and g77 could you create a link "ln -s gFortran g77". Would this allow his programs to compile correctly.
The problem with this approach is that from all of the documentation I could find, gFortran is only a F90 and F95 compiler. So, is the gFortran compiler going to still compile F77 code, under the approach you describe above, or is the problem going to be much more difficult...
Thanks!!
Dave H
No. Fortran 95 is downward compatible with fortran 77 code and should compile f77 programs, subroutines and data blocks without a hitch. That said, there are a number of deprecated features, allowed in f77, that are really ugly (arithmetic if, common blocks) as they facilitate bad programming habits. It wouldn't be a bad idea to scrutinize this old code for those constructs and clean it up. Because what is said to be fortran 77 often is fortran 66 or even fortran IV (or Fortran 54 :-)). Kudos to Suse BTW for bringing f95 into the distro. Regards, -- Jos van Kan www.josvankan.tk
On August 28, 2005 4:24 am, Jos van Kan wrote:
No. Fortran 95 is downward compatible with fortran 77 code and should compile f77 programs, subroutines and data blocks without a hitch. That said, there are a number of deprecated features, allowed in f77, that are really ugly (arithmetic if, common blocks) as they facilitate bad programming habits. It wouldn't be a bad idea to scrutinize this old code for those constructs and clean it up. Because what is said to be fortran 77 often is fortran 66 or even fortran IV (or Fortran 54 :-)).
I started programming WATFOR before they added structure, is there support in linux for the card punch and reader? After Fortran I moved on to APL has anyone had luck getting Sharp APL to run? I gave it a try a while ago and had trouble with sax (sharp APL eXecutable). I found the documentation didn't match up with the files. -- Collector of vintage computers http://www.ncf.ca/~ba600 Machines to trade http://www.ncf.ca/~ba600/trade.html Open Source Weekend http://www.osw.ca
Mike wrote:
On August 28, 2005 4:24 am, Jos van Kan wrote:
No. Fortran 95 is downward compatible with fortran 77 code and should compile f77 programs, subroutines and data blocks without a hitch. That said, there are a number of deprecated features, allowed in f77, that are really ugly (arithmetic if, common blocks) as they facilitate bad programming habits. It wouldn't be a bad idea to scrutinize this old code for those constructs and clean it up. Because what is said to be fortran 77 often is fortran 66 or even fortran IV (or Fortran 54 :-)).
I started programming WATFOR before they added structure, is there support in linux for the card punch and reader?
My first Fortran experience was in a class, back when I was in grade 12. We had to fill in pencil mark cards, to submit our programs. Later, I took another Fortran class in college, where we used WATFIV. In this course however, we had actual IBM terminals connected to the mainframe. As is often the case, there was often a problem finding an unused working terminal. I got around that by going home and dialing in, using Procomm+ on DOS & my old XT clone. This also meant I could use a decent editor, instead of the horrible line editor on the mainframe. I also did some of my homework on a VAX 11/780 at work (I didn't have one at home. <g>). I still have that WATFIV text on my bookshelf. Also, it's close to twenty years since I've even seen a punch card.
On Sun, 2005-08-28 at 11:58 -0400, James Knott wrote:
Mike wrote:
On August 28, 2005 4:24 am, Jos van Kan wrote:
No. Fortran 95 is downward compatible with fortran 77 code and should compile f77 programs, subroutines and data blocks without a hitch. That said, there are a number of deprecated features, allowed in f77, that are really ugly (arithmetic if, common blocks) as they facilitate bad programming habits. It wouldn't be a bad idea to scrutinize this old code for those constructs and clean it up. Because what is said to be fortran 77 often is fortran 66 or even fortran IV (or Fortran 54 :-)).
I started programming WATFOR before they added structure, is there support in linux for the card punch and reader?
My first Fortran experience was in a class, back when I was in grade 12. We had to fill in pencil mark cards, to submit our programs. Later, I took another Fortran class in college, where we used WATFIV. In this course however, we had actual IBM terminals connected to the mainframe. As is often the case, there was often a problem finding an unused working terminal. I got around that by going home and dialing in, using Procomm+ on DOS & my old XT clone. This also meant I could use a decent editor, instead of the horrible line editor on the mainframe. I also did some of my homework on a VAX 11/780 at work (I didn't have one at home. <g>). I still have that WATFIV text on my bookshelf. Also, it's close to twenty years since I've even seen a punch card.
It's been about that long since I've seen punch tape for CNC machine programs, thank G-d that the memory size was limited to less than 640K. :) That tape would have been massive.
Mike McMullin wrote:
It's been about that long since I've seen punch tape for CNC machine programs, thank G-d that the memory size was limited to less than 640K. :) That tape would have been massive.
I used to maintain some Data General Nova 800 systems. The disks in those systems were loaded from paper tape. It took the best part of an hour, at 10 char/sec on an ASR33 Teletype.
Jos van Kan wrote:
David Henderson wrote:
Donn:
If the only change is gFortran and g77 could you create a link "ln -s gFortran g77". Would this allow his programs to compile correctly.
The problem with this approach is that from all of the documentation I could find, gFortran is only a F90 and F95 compiler. So, is the gFortran compiler going to still compile F77 code, under the approach you describe above, or is the problem going to be much more difficult...
Thanks!!
Dave H
No. Fortran 95 is downward compatible with fortran 77 code and should compile f77 programs, subroutines and data blocks without a hitch. That said, there are a number of deprecated features, allowed in f77, that are really ugly (arithmetic if, common blocks) as they facilitate bad programming habits. It wouldn't be a bad idea to scrutinize this old code for those constructs and clean it up. Because what is said to be fortran 77 often is fortran 66 or even fortran IV (or Fortran 54 :-)).
Kudos to Suse BTW for bringing f95 into the distro.
Regards,
Here's a work around that I have just tested by successfully compiling the Tinker program (http://dasher.wustl.edu/tinker). I installed g95 and edited the tinker make files to reflect the change from g77 to g95. Worked perfectly. Of course one could install a soft link or an alias, rather than editing make files. -- Stephen P. Molnar, Ph.D. Life is a fuzzy set Foundation for Chemistry Stochastic and multivariant http://www.geocities.com/FoundationForChemistry
David Henderson wrote:
I have a question regarding the switch from gcc3.x to gcc4.x in OpenSuSE 10B3. I recently made this switch to fix a problem with my X server, but had to revert back to 9.3 because there is no g77 for 10 as gcc4.x uses gFortran. This is going to cause major problems for people in statistics and engineering as we often use a lot (and I mean a LOT) of legacy code written in F77. For instance, try removing g77 and then installing R-2.1.1 (http://cran.r-project.org) from source. It will not work, because there is a lot of legacy fortran code used.
Does gFortran not compile Fortran77 code? I would have thought F95 would be backward compatible with F77, so compiling Fortran77 code with a F95 compiler should be fine. If you're talking about just the name of the executable, you could simply solve that by creating a symbolic link f77 -> gfortran. Other than, I would report this as a bug to the opensuse project. /Per Jessen, Zürich -- http://www.spamchek.com/freetrial - managed anti-spam and anti-virus solution. Sign up for your free 30-day trial now!
participants (9)
-
David Henderson -
Donn Washburn -
James Knott -
jdd sur free -
Jos van Kan -
Mike -
Mike McMullin -
Per Jessen -
Stephen P. Molnar