Date: Sun, 17 Jul 94 18:18:46 JST
From: Stephen Turnbull <turnbull AT shako DOT sk DOT tsukuba DOT ac DOT jp>
To: eliz AT is DOT elta DOT co DOT il
Cc: djgpp AT sun DOT soe DOT clarkson DOT edu
Subject: v 1.11 "bugs" to fix in 1.12 

Eli--
    Please send me your "suspicious features" list---sounds like it
may do 90% of the work on the FAQ I'm supposed to be compiling ;-)
    Um, sounds a lot like 'lint', now that I think about it.
    By the way, I'm not so systematic about it as you are; I'm not a
programmer by profession, so my activity is sporadic and rarely seems
to justify it.  But I do keep informal notes, and rarely get bitten
that badly twice.  However, lots of folks are newbies or the
equivalent, trying hard and deserving help.  I think it makes sense to
make things "fail safe" (ie, before they get into production and
preferably before they're even run), rather than produce trivial (but
sometimes very confusing) failures once you're actually using
them.
    Re your proposed revision of fork():  Sounds a lot like what Marty
Leisner is advocating.  I wonder if it might be possible to organize
these things into a set of libraries that do more or less the same
things:
(1) a library that assumes fork()/exec() should do spawn(), and
popen() should create a temporary file, then do a system() and
redirect stdin from the file, and so on;
(2) a library that abort()s all those calls (DJ's proposal);
(3) a DV/X library that does process control and IPC "right";
(4) ditto, OS/2;
...etc.

Obviously, this is inappropriate for 1.12, and would require lots of
help from third parties (ie, not the core group of DJ and so on).
What do y'all think?
    --Steve