This is the mail archive of the
mailing list for the binutils project.
Re: PATCH: support for NEC SX architecture
- From: Jaka Močnik <jaka at xlab dot si>
- To: Dave Korn <dave dot korn dot cygwin at googlemail dot com>
- Cc: binutils at sourceware dot org, Erich Focht <efocht at hpce dot nec dot com>
- Date: Mon, 06 Jul 2009 11:33:46 +0200
- Subject: Re: PATCH: support for NEC SX architecture
- References: <firstname.lastname@example.org> <4A3CFFF7.email@example.com> <4A4111B9.firstname.lastname@example.org>
first of all, sorry for the late reply: I was away, sailing, until this
On Tue, 2009-06-23 at 18:32 +0100, Dave Korn wrote:
> Dave Korn wrote:
> >> bfd/ChangeLog
> > Look, I've kept you waiting long enough for this as it is; I'll send this
> > now and follow-up with the review of the BFD part shortly.
> I'm working on that now, but I've got one overall question I'd like to ask
> you separately about the SYMNMLEN change to BFD. It's largely mechanical,
> sure, but it's still very intrusive. I wondered if you had at any point
> considered the approach of making SX COFF a fully-fledged derived sub-class of
> plain COFF in the same way as XCOFF and ECOFF are? Wouldn't you just need to
> override the definition of SYMNMLEN with your own value, then you could reuse
> all the standard coff code, and end up with a much simpler patch overall?
well, the answer is twofold...
first of all the SX COFF flavour did not seem that much different when
compared to ordinary COFF to require another subvariant along the lines
of [E|X]COFF which would only be used on a single architecture.
however, the main reason was that my change (which is in the end only
exposed to BFD users by addition of bfd_coff_symnmlen(abfd) to "public"
BFD API) looked quite similar to the already existing
bfd_coff_filnmlen(abfd) and similar functions.