FORTRAN lint tool - toolpack1.2
Russell J. Allgor
toolpack1.2 that will compile on an HP 9000 system. I have
found versions in the Linux subtrees on various ftp servers,
and from perelandra.cms.udel.edu. The Linux version says
it is based on an HPUX port performed by Roman Augustyn
on September 22, 1993 for use with HPUX 8.0x
(tested on a HP 9000/730 under HPUX 8.05).
Does anyone have a version of this software that will run
under HPUX 9.x, or access to the HPUX 8.x code before it
was modified for Linux? I can probably modify the Linux
version to get it to run, but this will take some time.
If anyone has any other favorite lint checkers that aren't
mentioned in the FAQ, I'd appreciate the information. We currently
use the HP lintfor tool, and ftnchek.
Any help would be greatly appreciated.
Russell Allgor
MIT - Dept. of Chemical Engineering
ral...@mit.edu
Prof. Loren Meissner
>
> If anyone has any other favorite lint checkers that aren't
> mentioned in the FAQ, I'd appreciate the information. We currently
> use the HP lintfor tool, and ftnchek.
See Fortran Journal, Jul/Aug 94 for a review of FLINT static F77 code
analyzer.
Fortran Forum Sept 94 (and earlier) lists FLINT and other F77 static
analysis tools.
Ref: http://www.fortran.com/fortran
The following are the main F77 code analysis tools listed in FForum:
1. FLINT "detects inconsistencies in common blocks and in argument
lists, non-portable code, and usage of variables. Produces a call tree
and a symbol table. A version for C language is also available.
Single-user license $3900 [for UNIX or VMS;] also available for many
other hardware platforms. Information Processing Techniques Corp;
1096 E. Meadow Cir; Palo Alto CA 94304; (415) 494-7500; FAX 494-2758"
2. "ftncheck" which you have ("mon...@mary.fordham.edu")
3. "ForCheck" -- email: "for...@rullf2.leidenuniv.nl" -- also supports
some F90 features; available for PC (price: 1050 dfl), VMS, Primos,
MVS, etc.
4. At a different level, "VAST/77to90" does a lot of checking and code
restructuring while converting from F77 to F90. From Pacific-Sierra
Research for around $1000 depending on platform (discounts available).
(310) 314-2300; email: "in...@psrv.com"
-Loren Meissner
Prof. Loren Meissner
> 3. "ForCheck" -- email: "for...@rullf2.leidenuniv.nl" -- also supports
> some F90 features; available for PC (price: 1050 dfl)
Can somebody translate 1050 dfl (Dutch Florins?0 into dollars
(approximately)?
-Loren Meissner
Charles Stump
>"Russell J. Allgor" <rallgor> wrote:
>>
>> If anyone has any other favorite lint checkers that aren't
>> mentioned in the FAQ, I'd appreciate the information. We currently
>> use the HP lintfor tool, and ftnchek.
>
>See Fortran Journal, Jul/Aug 94 for a review of FLINT static F77 code
>analyzer.
>
>Fortran Forum Sept 94 (and earlier) lists FLINT and other F77 static
>analysis tools.
>
>Ref: http://www.fortran.com/fortran
>
>The following are the main F77 code analysis tools listed in FForum:
>
>1. FLINT "detects inconsistencies in common blocks and in argument
>lists, non-portable code, and usage of variables. Produces a call tree
>and a symbol table. A version for C language is also available.
>Single-user license $3900 [for UNIX or VMS;] also available for many
>other hardware platforms. Information Processing Techniques Corp;
>1096 E. Meadow Cir; Palo Alto CA 94304; (415) 494-7500; FAX 494-2758"
>
>2. "ftncheck" which you have ("mon...@mary.fordham.edu")
>
>MVS, etc.
>
>4. At a different level, "VAST/77to90" does a lot of checking and code
>restructuring while converting from F77 to F90. From Pacific-Sierra
>Research for around $1000 depending on platform (discounts available).
>(310) 314-2300; email: "in...@psrv.com"
>
>-Loren Meissner
>
You should also check out Refine/Fortran. It provides flow charts,
structure charts, data flow diagrams, cross references, and coding
standards checking. Info available at:
---
=======================================================================
= Charles Stump, Director of Business Development, cst...@levtech.com =
=======================================================================
= Leverage Technologists, Inc. = Reverse Engineering, =
= 6701 Democracy Blvd. = Reengineering, =
= Bethesda, MD 20817 = Maintenance, and =
= Suite 324 = Quality Assurance =
= (301)309-8783 = tools and services. =
= http://stout.levtech.com/ = =
=======================================================================
Dave Love
language restrictions that Toolpack (and NAGware -- which should be on
that list) do. I'm thinking specifically of storage association
restrictions and common save status.
[NAG's free f90 frontend (in the Market somewhere) does a good deal of
code checking and they state, as I recall, that you don't need other
static analysis tools because the compiler does enough, but it doesn't
seem to in that area.]
A potentially-valuable -- but rather noisy -- facility from g77 -O
-Wall that I don't remember seeing elsewhere is the `might be used
uninitialized' warnings you get from case analyses with holes in;
although such code is usually legitimate, it's not always.
John Appleyard
rallgor "Russell J. Allgor" writes:
> If anyone has any other favorite lint checkers that aren't
> mentioned in the FAQ, I'd appreciate the information. We currently
> use the HP lintfor tool, and ftnchek.
>
> Any help would be greatly appreciated.
>
plusFORT includes (among other things) an excellent Static Analyser which has
a number of nice features missing in others. See our new www page for details
(see below - still under construction in places).
--
John Appleyard Jo...@polyhdrn.demon.co.uk
Polyhedron Software Ltd.
Programs for Programmers - QA, Compilers, Graphics
************ Visit our Web site on http://www.polyhedron.co.uk/ ************
Thomas Koenig
>A potentially-valuable -- but rather noisy -- facility from g77 -O
>-Wall that I don't remember seeing elsewhere is the `might be used
>uninitialized' warnings you get from case analyses with holes in;
>although such code is usually legitimate, it's not always.
You might try the combination of f2c and checkergcc (or has anybody
integrated g77 with the checker-assembler yet?). It's devastatingly
effective at finding misbehaviour, especially uninitialized variables.
Unfortunately, checkergcc is only available for Linux so far.
--
Thomas König, Thomas...@ciw.uni-karlsruhe.de, ig...@dkauni2.bitnet.
The joy of engineering is to find a straight line on a double
logarithmic diagram.
Mr. P. Booth
>In comp.lang.fortran, Dave Love <d.l...@dl.ac.uk> wrote:
>
>>A potentially-valuable -- but rather noisy -- facility from g77 -O
>>-Wall that I don't remember seeing elsewhere is the `might be used
>>uninitialized' warnings you get from case analyses with holes in;
>>although such code is usually legitimate, it's not always.
>
>You might try the combination of f2c and checkergcc (or has anybody
>integrated g77 with the checker-assembler yet?). It's devastatingly
>effective at finding misbehaviour, especially uninitialized variables.
>
>Unfortunately, checkergcc is only available for Linux so far.
Is checkergcc a part of gcc v 2.7.0?
I can't find it using archie and haven't heard of it before.
Peter
------
Peter Booth Email: P.B...@ic.ac.uk
Imperial College Centre for Environmental Technology
48 Prince's Gardens, London SW7 2PE
0171-594 9296 (ph)
0171-581 0245 (FAX)
WoodPusher
I have a function to which is passed
a variable of INTEGER*2 (call it J).
I later pass J+1 to the same function.
J+1 is automatically promoted to an INTEGER*4,
thus causing this error listed in the subject header.
I would prefer that J+1 also be of INTEGER*2,
and if possible to do it without code intervention.
Is there a switch or compiler directive that will do this?
If not, what other options would you suggest?
Thanks,
WoodP...@aol.com
"To the victor go the spoils,
to the spoils go the bacteria.
Maybe we should call it a draw?"
Kurt Kallblad
>
>I am using MS-FORTRAN 5.1
>I have a function to which is passed
>a variable of INTEGER*2 (call it J).
>I later pass J+1 to the same function.
>J+1 is automatically promoted to an INTEGER*4,
>thus causing this error listed in the subject header.
>I would prefer that J+1 also be of INTEGER*2,
>and if possible to do it without code intervention.
>Is there a switch or compiler directive that will do this?
>If not, what other options would you suggest?
The metacommand "$storage:2" at the top of the source file will force
all integers (and logicals) not declared integer*4 to be integer*2.
Also expressions like j+1 will be of length 2. But using this
metacommand may lead to new problems with subroutines having formal
arguments declared as integer*4 .
Another solution is to ignore the error massage from the compiler. It
may be read as a warning as its workes as long as the actual argument
is within the numerical rage for integer*2. At least with MS-Fortran
5.0 the object code is produced, the linker does not complain and the
program works as it should. But this might not be a good "programming
style".
If none of these solutions work, I think you must go into the code and
use things like "INT2(J+1)".
Hope this will help.
Have a nice day,
Kurt
--
Kurt Kaellblad E-mail: kurt.k...@bkl.lth.se
Department of Building Science Phone: +46 46 222 7358
Lund Institute of Technology Fax: +46 46 222 4719
P.O.Box 118, LUND
S-221 00 Sweden
<Speaking for myself, not the department, nor ...>