Re: Platforms using COFF?

[Michael Snyder <msnyder@vmware.com>]

Eli Zaretskii wrote:
> I'm sorry for posting a slightly off-topic question, but I think this
> forum has many people who know about what I'd like to ask: are there
> any current or future platforms that use, or can be reasonably
> expected to use, the COFF format for their binary files?
> 
> Background: Emacs has a bunch of implementation of unexec, one each
> for every ABI used by platforms that Emacs can be built on.  (unexec
> is a method of generating an executable binary file from an in-memory
> image of a running program.)  There are, for example, unexelf.c for
> ELF, unexw32.c for pe-coff, etc.  There's also unexcoff.c (renamed
> yesterday from unexec.c), which was the original implementation
> supporting a.out and COFF.  Nowadays, it is used only by the MSDOS
> (a.k.a. DJGPP) build of Emacs.
> 
> The main issue is this: since only the MSDOS build uses unexcoff.c, I
> was asked to clean it up of code that is #ifdef'ed away for the DOS
> build.  My question is: can we reasonably assume that no future
> platform where it makes sense to build Emacs will ever want to use
> COFF?  If some modern low-end platforms (e.g., some mobile ones) use
> or could use COFF, then perhaps some of the code currently unused by
> the DOS build should be left in unexcoff.c, for the benefit of these
> platforms.

Not sure how meaningful, but...


grep -li coff config/*/*
config/djgpp/fnchange.lst
config/powerpc/aix.mh
config/rs6000/nm-rs6000.h