Date: Wed, 9 Apr 1997 18:34:08 -0400 (EDT) From: Pierre Phaneuf Reply-To: pierre AT tycho DOT com To: opendos-developer AT delorie DOT com Subject: Re: The compiling tools In-Reply-To: <860607233.0613892.0@abwillms.demon.co.uk> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII On Wed, 9 Apr 1997, Alaric B. Williams wrote: > > I'd say that from the moment it runs ok with the LFN support from DJGPP, > > we'll have something... > > Indeed. DJGPP can start by using the real mode interface, but I think > everyone will be eager to change libc to use an enhanced protected > mode interface to this TSR if one is found; that'll be faster. Yes, but for a primary release, striving for a 32-bit version of OpenDOS is asking quite a thing... As some said, probably will come as OpenDOS for the first release, then a subsequent OpenDOS/32 or something like that... To everyone in power, I *urge* them to take a look at TSX-Lite (available at http://www.sandh.com/__bbs__/tsxlite.htm ), a demo version of the commercial (and heftily priced) TSX-32. Personally, *this* is about what I need, maybe with some simple extensions on top of it (LFN)... > > I run Linux here, and when I do DOS or Win95, I do > > it with bash, using sh-utils (ls, mv, rm, etc...)... It's a shame they > > works ultra-nice in Win95, but stalls on the long or weird filenames in > > bare DOS (like choking on dotfiles)... > > With an ext2 IFS, you can use the same filesystem as with Linux :-) Yes, but anyway you have to use DOS binaries, DOS compilers and everything... My biggest project is build around a 16 bit real mode Modula-2 compiler... Gotta be DOS! As I said, for a primary release, simple API compatibility with the real mode interface to LFN would be more than enough in this regard... Thinking more about DOS extenders: not being a "high wizard", is there any reason (beside increased binary size) instead of emulating DOS calls by switching to real mode and calling the DOS kernel to do the IO, a DOS extender wouldn't have its own FAT (and others services) routines implemented as self-contained 32-bit procedures? With the disclosure of the OpenDOS sources, we could probably extract the procedures from there and convert those to 32-bit clean C (from ASM I guess)... Pierre Phaneuf