Native code BCPL compiler for Amiga?

[steve_mynott]

Originally Posted by mark_k

In makefile is this line:

.b.obj:; bcpl $*.b $*.obj opt '"%$$AMIDOS,$$AMIGA,$(BOPTS)"' obj A

Is the % symbol a typo? Also do the $$ get changed by make when invoking bcpl? (E.g. changing $$ to $ maybe??? Obviously I'm not an expert on this.)

I suspect the % does something although I'm not sure what!

A guess might be that "%" is something like the -D option to C?

Yes the doubled up $$ are basically escaped so they are passed to the UNIX shell by make correctly as a single $.

When I type "make" on a UNIX-type system I see something like

start.b start.obj opt '"%$AMIDOS,$AMIGA,$JUMPS,$MACRO,$EXT,$WORD,w30000"' obj A

So the $(BOPTS) gets evaluated out. The BCPL itself often uses $$ as well I notice.

I suspect the $AMIGA is used in the conditionals inside the BCPL which are like "$<AMIGA" (similar to C preprocessor?)

I've had a look online and can't find any docs.

There is a MMC QL BCPL kit (which might have full docs) on EBay but I was outbid!

[mark_k]

There's a PDF of the QL BCPL manual mentioned above. The book "BCPL: The Language and its Compiler" could perhaps contain some info, though of course it wouldn't be specific to the Metacomco compiler.

[mark_k]

It could be that % was a typo; it's Shift-5 for % vs Shift-4 for $ on the keyboard, and the corresponding UNIX opt line doesn't have a % in.

I don't know whether removing the % will make any difference, but I might try building the Amiga executable in NetBSD without it.

The BCPL manual has a section on conditional compilation on page 13 (p.21 of the PDF).

[modrobert]

Thinking of ancient Unix variants and their shells; the %$AMIDOS part is probably an environment variable, maybe it's supposed to point to sys:c on Amiga? Just a thought...

[Locutus]

I cant imagine that the % has anything to do with the UNIX world side of things, in ancient UNIX shells the % was used as a indicator for job numbers.

Providing as a opt the job number stored in a variable (i'm not even sure bsh/csh/sh from back then would evaluate that) as an argument sounds extremely unlikely.

Also seeing its escaped in the make output i'm pretty sure its passed verbatim to the BCPL side and probably only evaluated and expanded there by a preprocessor or the compiler. I do think the hunch that its some directives and/or variables for that.

[asymetrix]

Is BCPL useful today ? In this non Amiga post they chat about using a tiny FORTH program to 'bootstrap' other software. https://github.com/Fedjmike/mini-c/issues/2

[Locutus]

There's not really any 'modern use' of BCPL, Forth has some extreme niche uses as its one of the simplest languages you can imagine to implement a interpreter for and thats why its mentioned in that discussion.

[nogginthenog]

The obvious use would be to compile the TripOS bits of AmigaOS when it is eventually open sourced
(assuming there's any left in OS3.x)

[PurpleMelbourne]

So dragging up this old thread. I'm really curious to know how this went.

Has it been possible to compile, and port away from the original BCPL code?