Date: Tue, 10 Sep 1996 14:03:05 +0200 (IST) From: Eli Zaretskii To: DJ Delorie Cc: djgpp-workers AT delorie DOT com Subject: Re: V2 alpha 980908 -- djdev and djlsr In-Reply-To: <199609090348.XAA28882@delorie.com> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII 1) DJGPP_MINOR on lib/specs is still 0, not 1. 2) What about making libc_p.a (and maybe also libc_g.a) in a separate zip, so whoever needs them, could have them without rebuilding the entire library? Is that planned in the final release? 3) bnu27b.zip is larger than 1.44MB. Isn't that a constraint? If not, maybe I should change the packaging of Emacs too (someone who tried it told me that for people who have a direct net connection, it's a pain in the lower back to download 8 files instead of 1 or 2). 4) The new `write' doesn't set errno to ENOSPC when it cannot write all the bytes it was asked. I think it should. 5) crt1 now sets segment limit for _dos_ds to -1. Would this work for all the DPMI hosts? What about OS/2, DOSEmu, and others that take memory protection at face value? (And btw, why is this change needed?) 6) The fragment below is from `glob.c'. I think it's wrong to call `add(pathbuf)' if the file by that name doesn't exist. This should be only done if GLOB_NOCHECK flag is set, and the top-level code already does that when `glob2' returns with no files found. Right now, `glob' always returns with the original string, even if it didn't find any matching file and even if GLOB_NOCHECK is NOT set. if (*pp == 0) /* end of pattern? */ { if (__file_exists(pathbuf)) { if (flags & GLOB_MARK) { struct ffblk _ff; findfirst(pathbuf, &_ff, FA_RDONLY|FA_SYSTEM|FA_DIREC|FA_ARCH); if (_ff.ff_attrib & FA_DIREC) { char *_pathbuf = pathbuf + strlen(pathbuf); *_pathbuf++ = '/'; *_pathbuf = 0; } } if (add(pathbuf)) return GLOB_NOSPACE; } else if (wildcard_nesting==0) add(pathbuf); return 0;