Sender: bill AT taniwha DOT tssc DOT co DOT nz Message-ID: <36F69D7F.25F23B74@taniwha.tssc.co.nz> Date: Tue, 23 Mar 1999 07:43:59 +1200 From: Bill Currie X-Mailer: Mozilla 4.05 [en] (X11; I; Linux 2.2.3 i486) MIME-Version: 1.0 To: djgpp-workers AT delorie DOT com Subject: Re: djasm shift semi bug + proposal References: <36F5CF26 DOT AEED9A67 AT taniwha DOT tssc DOT co DOT nz> <199903221509 DOT KAA00687 AT envy DOT delorie DOT com> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Reply-To: djgpp-workers AT delorie DOT com DJ Delorie wrote: > > Please make sure nothing in djgpp uses sh[rl]d before changing the > meaning of that opcode. If you do that, I'll check in your patch. src/stub/stub.asm shr/shl only tests/libc/crt0/teststub.asm shr/shl only These are the only *.asm files in djgpp that use any sh[rl]* instructions. (find . -name '*.asm') in the cvs tree. > djasm can have its own directory when it has complete documentation > and omf support, but not before 2.03. Sounds like a good deal to me. I was expecting someting link 3.00:) > o finish omf support (do you still want that, DJ?) > > Yup. It should work with djlink :-) Thought so, though I'd been under the impression you've given up on djlink (hence the question). > o better symbol handling (allow forward references in most expressions) > > In what cases don't they work now? I can't remember the exact cases right now, I'll have to produce some test cases, but I remember being frustrated by this on occasion. > o better section support (first stage, being able to swap between > .text, .data and .bss at will, later maybe `real' segment support (esp > 16/32 bit control)) > > for coff this would be useful, but I think you're reinventing the > wheel here. Both gas and nasm already handle those kinds of programs. > But if you want to, and it doesn't destabilize it for djgpp, go ahead. But look at all those different wheels out there:) Seriously, `real' sements is probably way off (if ever), but a.out standard sections will be a real boon. Also, stability is very important to me (eg, the patch I sent is about 6 months old), so I have no intention of breaking stub.asm (in fact, I use it as a stability test). > o finalise instruction support for 486 > o pentium* instructions > > No problem here. So long as I don't get any more name clashes. If I do, I'll post hear to discuss what to do. Bill -- Leave others their otherness.