From: Chris G. Demetriou (cgd@gaia.CS.Berkeley.EDU)
Date: 02/22/93


From: cgd@gaia.CS.Berkeley.EDU (Chris G. Demetriou)
Subject: Re: Porting MH-6.8 to Linux....
Date: 22 Feb 1993 02:12:47

In article <1993Feb21.203127.5000@serval.net.wsu.edu> hlu@eecs.wsu.edu (H.J. Lu) writes:
=>|>
=>|> 1) Has anyone done the said port, or alternatively, were
=>|> similar problems faced for other OS's and how they were ironed out.

as a suggestion, take a look at the port of MH 6.8 to 386bsd.
it's available from agate.berkeley.edu:pub/386BSD/0.1-ports/somethingorother

we had to solve the same problems.

>MH-6.8 is unportable. You have to rewrite that part using stdio only.
>Not all systems have the same FILE. Take a look at HP-UX 8.0.x.

agreed, but it can be solved with a few well-placed #define's, in most
places...

writing it in straight stdio could slow MH down significantly...

>No. MH-6.8 tries to skip stdio and talk to FILE directly. No way you can
>do that on all systems. That part of MH-6.8 is badly coded.

to the latter, i say "bullshit," even though i was the first to get
an mh 6.7.* port completely working under 386bsd.

"badly coded" implies without justification. There's sufficiently
good reasoning as to why that chunk of MH is coded that way -- read
the comments, and you'll find out...

VJ might be a speed demon, but he's smart about it; he finds out
what needs to be optimized, then does it...

chris
who thinks the fact that MH uses the internals of FILE is ugly, too