riscosify() and guess_or_null()

John Tytgat John.Tytgat at aaug.net
Mon Sep 29 14:10:33 PDT 2003


I just got a bug report from a CVS user which boiled down to a bit of
heuristic code in the UnixLib riscosify() routines.  I.e. the code tries
to figure out if the incoming filespec is a Unix or RISC OS file spec.
Normally you would expect Unix filespecs but in case of e.g. "foo.c.getopt"
is recognised as a RISC OS filespec (so copy_or_null() is called) because
"c" is in the sfix list.

Although the user is helped with a careful sfix choice for his particular
CVS project, the analysis of the problem is far from logical (why is
".new.msg.1.f.v" is Unix filespec and "msg.1.f.v" a RISC OS filespec ?
I'm taking a real life example here).

So I would like to ask if there is any objection to have a
__riscosify_control bit allocated which, if set, redirects guess_or_null()
straight to translate_or_null().  Default I would leave it unset, so that
current heuristic behaviour remains as it is.

Objections ?

John Tytgat, in his comfy chair at home                                 BASS
John.Tytgat at aaug.net                             ARM powered, RISC OS driven

More information about the gcc mailing list