From what has been said about SuperSoft FORTRAN (here and elsewhere) I had assumed that it was a fairly high quality piece of software. I've downloaded a zip file, and tried to use it. I can't get anything I've written to work. The zip file included a little "test" program, and I can get that to compile and run, BUT the structure of the program is somewhat unorthodox (IMO). Even a simple loop program, that I wrote, gives run-time errors (and the documentation that says to look in Appendix G(?) for information about run-time errors, has no Appendix G(?) or any appendices). Anybody have any experience with SuperSoft FORTRAN? What am I doing wrong? As an aside, I wrote tons of FORTRAN code in graduate school, so I'm not a FORTRAN newbie.. Is it just crippleware??

In contrast, I have been able to use the Microsoft FORTRAN software with only limited hassles.

Roger

суббота, 15 января 2022 г. в 22:59:52 UTC+3, norwe...@gmail.com:
> From what has been said about SuperSoft FORTRAN (here and elsewhere) I had assumed that it was a fairly high quality piece of software. I've downloaded a zip file, and tried to use it. I can't get anything I've written to work. The zip file included a little "test" program, and I can get that to compile and run, BUT the structure of the program is somewhat unorthodox (IMO). Even a simple loop program, that I wrote, gives run-time errors (and the documentation that says to look in Appendix G(?) for information about run-time errors, has no Appendix G(?) or any appendices). Anybody have any experience with SuperSoft FORTRAN? What am I doing wrong? As an aside, I wrote tons of FORTRAN code in graduate school, so I'm not a FORTRAN newbie. Is it just crippleware??
>
> In contrast, I have been able to use the Microsoft FORTRAN software with only limited hassles.
>
> Roger

I haven't done a full comparison of the Microsoft and SuperSoft FORTRAN compiler. However, both of these compilers mostly conform to the ANSI X3.9-1966 standard. Both have small extensions and differ from each other in some operators. The text of the program in the SuperSoft FORTRAN language should be written in capital letters. Microsoft FORTRAN does not have these limitations. For this reason, source files from Microsoft FORTRAN may not be compiled by SuperSoft FORTRAN.
I advise you to read the SuperSoft FORTRAN manual. It's not complete, but it describes the language well enough that it can compile. The SuperSoft FORTRAN compiler is pretty good, But it requires care, As they say, you can hammer nails with a hammer, but at the same time beat yourself on your fingers, From this the hammer does not become a malicious object.

On Saturday, January 15, 2022 at 3:46:03 PM UTC-8, nikiti...@gmail.com wrote:

....<snip>....
>> The text of the program in the SuperSoft FORTRAN language should be written in capital letters.

Right -- I gathered that from looking at the examples in the documentation.

>> For this reason, source files from Microsoft FORTRAN may not be compiled by SuperSoft FORTRAN.

I never expected that to be the case. I've fooled around with enough CP/M compilers that I know that they vary widely in their operational requirements.

>> It's (the manual) not complete, but it describes the language well enough that it can compile.

I just wish that the appendices were available somewhere. The software just spits out numbered error messages which aren't very useful to figure out exactly what's gone wrong.

I discovered (quite by accident -- I downloaded the files to my Z80 via XMODEM on my B serial port, and left the port hooked up after doing so) that a lot of feedback from the compiler and utilities gets sent to that port. I guess that's the CP/M "printer port"?. That helped me figure out what I was doing wrong, and I finally got a simple DO loop program to work as expected. Maybe now I can get something more complex to work?

Thanks for your comments.

Roger

понедельник, 17 января 2022 г. в 01:35:06 UTC+3, norwe....@gmail.com:
> On Saturday, January 15, 2022 at 3:46:03 PM UTC-8, nikiti...@gmail.com wrote:
>
> ...<snip>....
> >> The text of the program in the SuperSoft FORTRAN language should be written in capital letters.
> Right -- I gathered that from looking at the examples in the documentation.
> >> For this reason, source files from Microsoft FORTRAN may not be compiled by SuperSoft FORTRAN.
> I never expected that to be the case. I've fooled around with enough CP/M compilers that I know that they vary widely in their operational requirements.
>
> >> It's (the manual) not complete, but it describes the language well enough that it can compile.
>
> I just wish that the appendices were available somewhere. The software just spits out numbered error messages which aren't very useful to figure out exactly what's gone wrong.
>
> I discovered (quite by accident -- I downloaded the files to my Z80 via XMODEM on my B serial port, and left the port hooked up after doing so) that a lot of feedback from the compiler and utilities gets sent to that port. I guess that's the CP/M "printer port"?. That helped me figure out what I was doing wrong, and I finally got a simple DO loop program to work as expected. Maybe now I can get something more complex to work?
>
> Thanks for your comments.
>
> Roger

I am glad that the problem has been fixed. I wish you success in working with this compiler.
I also want to find the full documentation.

Andrey

I was finally able to get the test program I wanted to run to compile. I found that execution with the SuperSoft version took 1 min. 16 secs (20 MHz Z80), but the Micro$lop FORTRAN version ran in 21 or 22 secs. The SuperSoft code is smaller (8.4k vs.12.8k). The program uses floating point arithmetic, so there were slight differences in the output, but I expected that. Results depend on how the f.p. is implemented. In debugging the SS code (and checking the numbers with a calculator), I think that the SS results are pretty accurate.

As a point of reference, the same algorithm compiled with Hitech C runs in about 16 seconds (same computer).

Interesting exercise!

Roger

On 18/01/2022 08:59, Roger Hanscom wrote:
> I was finally able to get the test program I wanted to run to compile. I found that execution with the SuperSoft version took 1 min. 16 secs (20 MHz Z80), but the Micro$lop FORTRAN version ran in 21 or 22 secs. The SuperSoft code is smaller (8.4k vs.12.8k). The program uses floating point arithmetic, so there were slight differences in the output, but I expected that. Results depend on how the f.p. is implemented. In debugging the SS code (and checking the numbers with a calculator), I think that the SS results are pretty accurate.
>
> As a point of reference, the same algorithm compiled with Hitech C runs in about 16 seconds (same computer).
>
> Interesting exercise!
>
> Roger

IIRC MS Fortran docs state it had one of the fastest 8080 f/p implementations.