Mail Archives: djgpp/1996/02/13/03:16:06
Hi everyone again| I'm back in the DJGPP mailing list. I got my UNIX account
closed for a month for some *personal* reasons. Anyway, who cares. Now I'm
writing from VMS (veeeeeehhhh.....)
This time I'm here to apologize for the promises I made about a new malloc()
for djgpp. I bumped my head into the wall... It's the sbrk()! I never thought
it could jump to a different DPMI memory block sometimes! i.e. sometimes
sbrk(1024) could move the break ahead for far more than 1024 bytes. Using
crt0 flags definetly is not an option for a thing tending to be a *library*...
OK I'm sorry, I'll work on that, but the whole algorythm of mine has to be
changed...
and now something more: I think very few people need both coff and exe files
after linking. For myself, I modified stubify.exe to *always* remove the coff
file and changed the specs file for stubify not being invoked if -noexe option
is specified. This makes gcc -o test test.c produce test.exe and no test. I
find that more comfortable.
Glad to be back...
Martynas Kunigelis
- Raw text -