Learning Fortran
in [Matlab]
|
From: Baalzamon on 20 May 2010 07:37 I am attempting to convert some old fortran codes and becoming somewhat annoyed by the apparent jumping of code. WOuld I be correct in assuming that the code is read top to bottom and in line number order? If so where does the do label numbers figure into this scheme? Any advice from fortan programmers would be appreciated. To be explicit I think the coxde was written in F77...
From: Rune Allnor on 20 May 2010 08:18 On 20 Mai, 13:37, "Baalzamon " <baalzamon_mori...(a)yahoo.com> wrote: > I am attempting to convert some old fortran codes and becoming somewhat annoyed by the apparent jumping of code. The infamous spaghetti style coding http://en.wikipedia.org/wiki/Spaghetti_code is the main reason why you ought not learn fortran at all. It was obsolete 40 years ago. The only reason it remains in use is the vast amount of legacy code present. > WOuld I be correct in assuming that the code is read top to bottom and in line number order? No. It *starts* that way, but from the first jump / break / continue / goto statement you are on your own with respect to untangling program flow. Hence the spaghetti metafor. > If so where does the do label numbers figure into this scheme? They are the spots at which a jump / break / continue / goto might refer. One serious problem with fortran is that there is no convenient way to find out where the jump that landed on the label was made. > Any advice from fortan programmers would be appreciated. To be explicit I think the coxde was written in F77... Don't write fortran. It might be more cost effective to find out what the program is supposed to do, and then develop and code the algorithm from scratch in matlab, C, C++ or Java. No kidding. Rune
From: Baalzamon on 20 May 2010 08:26 Ah right much thanks. I don't intend to actually learn fortran - or rather just enough to understand the darn code. Well i'll ontinue my attemts at following it but like you suggested i might just see if i can obtain the actual maths behind this code and attempt to implement it straight into matlab. Fun Fun Fun? Anyway thanks for the reply.
From: Rune Allnor on 20 May 2010 09:36 On 20 Mai, 14:26, "Baalzamon " <baalzamon_mori...(a)yahoo.com> wrote: > Ah right much thanks. I don't intend to actually learn fortran - or rather just enough to understand the darn code. Well i'll ontinue my attemts at following it but like you suggested i might just see if i can obtain the actual maths behind this code and attempt to implement it straight into matlab. Fun Fun Fun? Fact of Life. And a reason for paying accute attention to the algorithm development and documentation as opposed to the source code. Whenever I implement something, I write a companion document where all the minute deliberations are recorded. I might not use the first matlab version for eternity, but as long as I have access to that document I can implement the algorithm in whatever language in a matter of hours. Rune
From: dpb on 20 May 2010 09:39
Baalzamon wrote: > Ah right much thanks. I don't intend to actually learn fortran - or > rather just enough to understand the darn code. Well i'll ontinue my > attemts at following it but like you suggested i might just see if i can > obtain the actual maths behind this code and attempt to implement it > straight into matlab. Fun Fun Fun? > Anyway thanks for the reply. Fortran is _NOT_ obsolete; the last Standard was for F2003 and F2008 is in process. F77 was supplanted by F90/95 which incorporated much of dynamic memory and other features. Control for structured programming have been in common extensions since well before F77 and F77 formalized IF blocks while F90 standardized CASE and DO WHILE, etc. The answer to the specific questions you posed are... a) Yes, code is sequential until a branch path is taken b) Label numbers are only significant as the target for a branch (GOTO computed IF, DO construct end or ERR= or END= location, etc.). The magnitude of the label is not significant altho most coders had a general convention the value itself is immaterial to the structure of the code; it's just a label. One key to Fortran that sometimes puzzles those who come from a C background--the scope of Fortran compilation is the program unit, which is _NOT_ the file but the individual PROGRAM, FUNCTION or SUBROUTINE mor than one of which can be contained within a single file. Thus there are not "file scope" variables, etc. For F77, global variables, if any, typically will be in COMMON blocks (hopefully all of which will be named, not blank COMMON). There are tools that can help -- I've been very pleased w/ the demonstration/trial usage of the "Understand for XXXX_Language" analyzers from Scientific Toolworks <www.scitools.com> to help in digging into legacy code. At the last time I actually had need altho I've been retired for a few years now, it was still available fully functional for a reasonable time as a free download that was sufficient to accomplish a relatively small task such as yours sounds as if would be. I know there are some other open source tools but I don't have a handle on them; requests along that line would best be posted to comp.lang.fortran instead of c.s-s.m. Also, if have future specific questions regarding the code itself c.l.f is active and very helpful w/ several who are both ML and Fortran-literate. -- |