[comp.lang.fortran] On Microsoft Fortran 5.0. Bugs again?

bauwens@euler.Berkeley.EDU (Luc Bauwens) (06/29/89)

So I finally convinced my customer to convert his old BASIC stuff to
Fortran.  To make the conversion smoother, we stick with Microsoft,
and we decide to start converting the most computation-intensive
routine.

So, here we go, and buy the latest version of MS Fortran, 5.0 or
something like that.  I install it on his Everex 386/20.  

Testing runs OK, except the Microsoft editor gives some funny
messages right after being loaded, but apparently ends up
working OK.  So I start and convert a perhaps 300/400 lines
routine, and debug the conversion.

First time through works OK, produces the usual long error
list, mostly declaration problems.  Then the fun
starts...

When trying to compile, depending upon the flags specified, 
either the machine freezes, or starts a sequence of repeated
printscreens...   But compile, it doesn't...

So my question: am I doing something desperately wrong, or
is Microsoft really beyond salvation?

Luc Bauwens

silvert@cs.dal.ca (Bill Silvert) (06/30/89)

In article <15054@pasteur.Berkeley.EDU> bauwens@euler.Berkeley.EDU.UUCP (Luc Bauwens) writes:
>So, here we go, and buy the latest version of MS Fortran, 5.0 or
>something like that.  I install it on his Everex 386/20.  

If posting a bug report, how about getting the details straight?
I have 4.1, which I think is the current version.  How can we comment if
we don't know exactly what you are using?  Are you sure the machine
isn't a 486/50 or something like that?

>Testing runs OK, except the Microsoft editor gives some funny
>messages right after being loaded, but apparently ends up
>working OK.  So I start and convert a perhaps 300/400 lines
>routine, and debug the conversion.

What messages?  Maybe they mean something!

>When trying to compile, depending upon the flags specified, 
>either the machine freezes, or starts a sequence of repeated
>printscreens...   But compile, it doesn't...

What flags?  What command line are you using?  What are you doing?

>So my question: am I doing something desperately wrong, or
>is Microsoft really beyond salvation?

Since you haven't told us what you are doing, how can we tell whether
you are doing it right or wrong?

For the record, I have MS 4.1 and maintain a sizeable simulation
software package with it, including links to C and assembler routines.
It works fine, not cryptic messages or anything like that.  The code is
all written in standard Fortran-77 except for the include statement, and
I do all program development on a Unix system, then just run through sed
to convert the includes and compile on a Dell 310 (386/20 also).

The MS make is pitiful, and I haven't tried CodeView, since I don't
bother using MS-DOS for development.  I stick to Fortran-77, so
conversion is not a problem.  

If I do run into problems, I will notify this newsgroup, and I will
include the details.  Suggest others do likewise.

-- 
Bill Silvert, Habitat Ecology Division.
Bedford Institute of Oceanography, Dartmouth, NS, Canada B2Y 4A2
	UUCP: ...!{uunet,watmath}!dalcs!biomel!bill
	Internet: biomel@cs.dal.CA	BITNET: bs%dalcs@dalac.BITNET

bauwens@euler.Berkeley.EDU (Luc Bauwens) (06/30/89)

In article <1989Jun29.175743.28105@cs.dal.ca> bill@biomel.UUCP writes:
>In article <15054@pasteur.Berkeley.EDU> bauwens@euler.Berkeley.EDU.UUCP (Luc Bauwens) writes:
>>So, here we go, and buy the latest version of MS Fortran, 5.0 or
>>something like that.  I install it on his Everex 386/20.  
>
>If posting a bug report, how about getting the details straight?
>I have 4.1, which I think is the current version.  How can we comment if
>we don't know exactly what you are using?  Are you sure the machine
>isn't a 486/50 or something like that?
 
Read the sentence above :-).  I am pretty sure it was a version
5.something, and it must have been 5.0.  The computer is an Everex 386/20.

>>Testing runs OK, except the Microsoft editor gives some funny
>>messages right after being loaded, but apparently ends up
>>working OK.  So I start and convert a perhaps 300/400 lines
>>routine, and debug the conversion.
>
>What messages?  Maybe they mean something!

Sorry, but I did not note exactly.  They referred to the editor, M, and
not the compiler, and I doubt that they were relevant.  They mentioned
something like 'next file is ....'(I don't remember the name of the file).
This was followed by 'press return', and a number 100 appeared on the
bottom right side of the screen.  Then came a message '....is not an
editor command.  Press return', or something like that.  However
after that, the editor worked OK, for what I could see.

Anyway, I tried compiling from within and without the editor, and
the results were identical.  So I don't think the editor was 
involved.

>>When trying to compile, depending upon the flags specified, 
>>either the machine freezes, or starts a sequence of repeated
>>printscreens...   But compile, it doesn't...
 
>What flags?  What command line are you using?  What are you doing?

OK.  I used the command FL.

I tried without flags or with a flag turning optimization off 
(I think it was /Os or something like that).  It then would start
printing the screen repeatedly.

When I tried fl /c (produces an object file but doesn't try to link),
or /Zs (I think - this is from memory, and I don't have the manuals
here), which means only to check the source code without producing
anything, it would just freeze.

I tried also the flag which produces a source listing.  This didn't
do anything.

One more detail: I had installed the whole thing in the most
straightforward manner, nearest to the default thing.  For the
editor, there was a choice of customization, which I set at 'quick'.

>I do all program development on a Unix system, then just run through sed
>to convert the includes and compile on a Dell 310 (386/20 also).
 
Yes, perhaps that's what I ought to do also...  Except presently my
customer does not have access to a Unix machine.

>The MS make is pitiful, and I haven't tried CodeView, since I don't
>bother using MS-DOS for development.  I stick to Fortran-77, so
>conversion is not a problem.  

This Fortran 5.0 includes Codeview, and I may try it.  But if the
compiler dies on me, what's the use?

Luc Bauwens