GPC going going gone? Return of IP Pascal. What about FPC?
5th version of the Pascal-P compiler
Brought to you by:
samiam95124
Menu
▾
▴
GPC going going gone?
As you might have noticed, I spent a lot of time trying to get together a working version of GPC on each of the Windows, linux and Mac platforms.
GPC has a terminal disease that is not going to go away. Its incompatible with newer versions of GCC, and it needs GCC to compile programs. GCC is the backend to GPC, GPC "hitches a ride" on GCC. As someone on the GPC mailing list stated to one of the poor souls trying to cobble together a working version of GPC, ideally GPC must be matched to the GCC version it was originally compiled for.
Now I have noticed that this is not exactly true. I am currently running GPC 3.4.5 on a Mingw base GCC of 4.6.2. But it is true enough. Mismatch the versions and you don't know what you get. I have given up on getting GPC running on the Mac. I came close, given a lot of work, then found the recipie collapsed the next time they updated the Mac OS. And they do that a lot.
Thus I have returned to running Pascal-P5 with IP Pascal. This is not ideal, since nobody has access to that compiler[1]. I won't give up on GPC runs. On Windows it is not hard to arrange a working GPC in 32 bit mode. On Linux, unless you select a 32 bit OS, you are going to need a 64 bit version of Pascal-P5, which I have worked hard to arrange. I'll be testing that further as time permits.
All of this makes a self strapping Pascal-P an imperative (ie, a Pascal-P with its own code generator). This is non-ideal with pascal-P5, because it has none of the extentions a normal compiler has (file handling comes to mind). A pascal-P5 with code generation is still (to me) a pretty useless system, but it would solve the issue of a vanishing GPC, serve as a good example of an ISO 7185 starter kit, and serve as a good base for Pascal-P6.
Occasionally someone emails me about getting Pascal-P5 running on FPC. I even had a series of emails claiming that FPC, with ISO 7185 compatability mode enabled, would be able to compile and run Pascal-P5 without modification.
The short answer is: I'd love to see it. You don't need to fight to get Pascal-P5 running on FPC as a first step. Read the documents. There is a good acceptance test for ISO 7185 mode in iso7185pat.pas. Run that correctly, and compiling and running Pascal-P5 is next.
Realistically what we have seen is FPC inching closer to a true ISO 7185 mode, and Pascal-P5 needing fewer and fewer mods to get running on FPC. If and when FPC can run Pascal-P5, I'll be adding that option to the Pascal-P5 tooling. I'd be happy to.
[1] Occasionally I get inquiries about IP Pascal. I prepared a commercial version of that some while back. There just wasn't the interest to justify trying to sell it. The open source demand for ISO 7185 compilers was well covered by GPC, and will be covered by Pascal-P5 and Pascal-P6. One of the issues with GPC was that they were trying to compete with FPC. My take on that has always been clear: if you want the FPC dialect, use FPC.
Would you consider using P5c --> https://sourceforge.net/projects/pascal-p5c/ ?
It compiles p5 and pint, is freely available and runs on linux (in fact it should run wherever gcc runs)
Also, I am still actively working on it, so if there are any problems, I could take a look at them.
So yes I want to try that. I only just discovered your project, I was looking for the FPC modified version of P5 and there you were.
I assume it is not zero work, since you didn't have incentive to also compile pint.pas. But yes, I'd like to validate that as an alternate implementation path.
I looked at your project, and it looks very well done. Congratulations.
I can't think of anything better than the idea that others are able to use my work to build something else :-)