Thanks for the port! Quick question:

Since amigaos device names are no longer allowed, how should I modify makefiles that refer to eg SOMEASSIGN:somedir/somefile.c?

Quote:

Since amigaos device names are no longer allowed, how should I modify makefiles that refer to eg SOMEASSIGN:somedir/somefile.c?

Yes.

I'd recommend getting rid of such assignments if you can. If not, change it to a POSIX-style path (/SOMEASSIGN/somedir/somefile.c) and avoid using AmigaOS commands which cannot deal with POSIX-style paths (e.g. cp instead of Copy) or you'll get a surprise when you use $@ and similar.

BTW, if you can think of some way to parse out AmigaOS device names from construct such as this:
dev:dir:%.c:dev:dir/%.o (note no spaces)

We could put the feature back in. The problem is some tools which automagically create makefiles don't put spaces around the colon and break only on AmigaOS so it makes porting things that much more difficult. I could add the feature back into "make" and remove it for "gmake" as well but I thought it was a hack to start with (depended on optional spaces) so why not just get rid of it.

_________________
ExecSG Team Lead

@ssolie

I'm unfamiliar with that makefile construct. Are you sure you don't mean

dev:dir/%.c:dev:dir/%.o

In that case, a left-recursive grammar to parse that (and paths without device names) could be

dependency ::= left ":" right
left ::= pattern
right ::= pattern
pattern ::= device ":" dirs-opt file
pattern ::= dirs-opt file
device ::= "a valid amigaOS device name, minus the colon. eg [a-zA-Z0-9_]+"
dirs-opt ::= "/" dir more-dirs
dirs-opt ::= dir more-dirs
dirs-opt ::= ""
more-dirs ::= "/" dirs-opt
file ::= "a valid amigaOS filename, plus % for makefile subst"

Turning the grammar into actual code is left as an exercise to the reader (NB: someone check my work, it's been a while...)