So I basically said "interpreters are useless" in this message, and I would like to set the record straight.

There are actually two important uses, for interpreters in general and for Pascal-P in specific:

1. Error checking. Pascal-P5 does more checking of the ISO 7185 source code than any other compiler in existence since the "Model implementation of Pascal" [Welsh&Hay]. It does checks that would be impractical in a full compiler, like check if you are making changes to variant record tags when there is a VAR reference to same outstanding (an almost impossible check to do in machine code).

In fact, if my GPC compiled code crashes, I can often compile it using Pascal-P5 itself, run it interpretively, and it will find the error, politely stop, and tell me what occurred and what line it happened on, as opposed to crashing and burning.

2. Accurate debugging. This is more of a Pascal-P6 thing, but compilers and their associated debuggers tend to do a much worse job of debugging, especially source level, than interpreters. Optimization will rearrange the code so that it does not look like the source, and usually you have to turn off optimization to get reasonable results. Even beyond that, interpreters have a 1:1 relationship with both the generated code and the data in memory, and you just get better and more accurate results from debugging an interpreted version.

Scott Franco

--------- Original Message --------- Subject: RE: Re: Plan to Update GPC From: "scott andrew franco" samiam@moorecad.com Date: 8/13/20 8:02 am To: "Schneider, Thomas (NIH/NCI) [E]" schneidt@mail.nih.gov, "gpc@gnu.de" gpc@gnu.de

Well, you need a working ISO 7185 compiler. There are actually a couple of them. I'm not sure the Prospero compiler is still available, I used to use it. IP pascal works, but even I don't use it at the moment, and I wrote it. It needs updating. The GPC compiler works quite well, and I use both the 32 bit version and the 64 bit version (windows and Linux), although (as noted in the documentation) only a single release version that has shown stability for the ISO 7185 language.

For your second point, I assume you are talking about the Pascal-P1-P6 thing? (not sure I understand your point about "name change breaks the code")

Yes, that one is a bit odd. Realize the compiler is 47 years old. It went through "versions" P1 though P4 long before I got it. I suppose that Niklaus Wirth wanted to clearly differentiate the versions, which were incompatible with each other. The name Pascal-P5 was picked a long time ago 1970s or early 1980s in fact) as being the full and not subset version of Pascal-P, which never got actually done until I did it, which brought it to ISO 7185 status. Pascal-P5 does, in fact, have version numbers, I believe the last is 1.3. And yes, I perpetrated Pascal-P6, because it extends ISO 7185, and thus is quite different again.

GPC used to work on Mac OS, and I used to release Pascal-P5 on Mac (its 10 years old now!). At some point, I think well documented in this mailing list, it stopped working.

Is Pascal-P5 useful? That's a complex question. The purpose of Pascal-P5 was to be a model compiler, that is, be a model of how to implement the ISO 7185 language and not necessarily a stand-alone system in itself. It also has a large body of tests and is designed to replace (and indeed surpass) the BSI test suite for ISO 7185 Pascal, which is no longer available (and was paidware when it was).

Can Pascal-P be used on the Mac? Well, actually yes. Pascal-P6 includes sufficient tooling to port without GPC (see the interpreter written in C). I don't really push that capability because P6 is still in development, and also because interpreters are not really my thing (yes, I know that sounds funny). I like compilers. To me interpreters are more of a fun toy than not, although I note that there is a group pushing interpreters as "VM"s for professional use. Pascal-P5 has actually been written up as such. Again, not bad, just not my thing.

I should mention there is a branch from P5 known as P5c that that changed the compiler into a Pascal to C translator that accomplishes porting in a different way. You would have to look into that yourself, I'm not really up to speed on that.

Regards,

Scott Franco

--------- Original Message --------- Subject: Re: Plan to Update GPC From: "Schneider, Thomas (NIH/NCI) [E]" schneidt@mail.nih.gov Date: 8/13/20 6:43 am To: "gpc@gnu.de" gpc@gnu.de

Hmm, is this your work https://sourceforge.net/p/pascalp5/ ?

What is this? I looked into the README and it says:

"You need a working gpc implementation."

I'm looking for a replacement for GPC on macOS since GPC no longer works there. How would pascalp5 be useful for anything if it depends on GPC which doesn't work on macOS?

Also, why is the name of the program changing instead of a version number? This will break code everytime you change the name!

Tom

Thomas D. Schneider, Ph.D. Senior Investigator National Institutes of Health National Cancer Institute Center for Cancer Research RNA Biology Laboratory Biological Information Theory Group Frederick, Maryland 21702-1201 schneidt@mail.nih.gov https://alum.mit.edu/www/toms

_______________________________________________ Gpc mailing list Gpc@gnu.de https://www.g-n-u.de/mailman/listinfo/gpc _______________________________________________ Gpc mailing list Gpc@gnu.de https://www.g-n-u.de/mailman/listinfo/gpc