Newsgroups: comp.sys.mac.programmer
Path: utzoo!news-server.csri.toronto.edu!rpi!usenet
From: Garance_Drosehn@mts.rpi.edu
Subject: Re: MM Init
Message-ID: <QL+=#F_@rpi.edu>
Nntp-Posting-Host: gilead.its.rpi.edu
Organization: Rensselaer Polytechnic Institute
References: <4017@uakari.primate.wisc.edu> <1991Mar11.173337.1044@ni.umd.edu> <565@genco.bungi.com> <50342@apple.Apple.COM>
Date: 18 Mar 91 16:50:32 GMT
Lines: 69

>In article <565@genco.bungi.com> rad@genco.bungi.com (Bob Daniel) wrote:
>Has the IIci ROMs been updated in newer IIci's so that the MM Init is no
>longer needed?  If so, what version of ROM has the fix?  A friend of mine
>just got a new IIci and was wondering if he needs to use the init.
 
In article <50342@apple.Apple.COM> keith@Apple.COM (Keith Rollin) replied:
>This is a copy of the letter I sent to Bob:
>
>Bob,
>
>IMHO, almost no one should need to be using MMInit. This is for several
>reasons which I'll list below, but first let me answer your question.
>No, the fix to the Memory Manager is not in any newer ROMs (Apple can't
>just quickly slap out new ROMs every day, you know), but a patch to the
>Memory Manager _is_ is System 7.0.
>
>As for the reasons why you shouldn't need to run MMInit:
>
>1) Almost no applications are affected by the bug. There is only one
>   that we know about, ProGraph. If you are not running that, then it
>   is 99.99% likely that running the INIT will have no affect on your
>   IIci.

Another application where a noticable speedup occurs is the one I'm using right
now, namely, TheNews.  The improvement is mainly noticable when going thru
newsgroups with a lot of articles in it.

>2) The INIT is a duplicate of the Memory Manager in the IIci ROM, with
>   the bug fixed. This duplicate includes other bugs in the Memory
>   Manager that occur only in 32-bit mode. These problems are fixed in
>   System 7.0 (which is the only way you can run in 32-bit mode), but
>   if you run the INIT, you'll re-introduce those bugs into your Mac.

Correct me if I'm wrong, but it sounds like this reason should be reworded as
"Don't use MMInit if you're running any system later than 6.0.7".  The fact
that bugs are fixed in 7.0 isn't going to mean much to me, running 6.0.5 on a
Mac IIci.  Remember, officially speaking I can't run system 7.0 even if I
wanted to (not yet, that is...).

> ...(skipping along)...
>4) The INIT has shown some compatibility problems with some major
>   3rd party software. The symptoms of these problems include crashes
>   into the debugger, mouse freezes, and the inability to boot your
>   Macintosh.

I must admit I have had some problems with system hangs.  Many of them went
away when I switched MMInit to be close to the first init loaded, but I still
had some problems.  I haven't tracked down the exact culprit(s) yet, but I
suspect something in my INITs and cdevs.  I had over 30 after I added MMInit,
and the remainder of my hanging problems went away when I disabled a couple of
those.  Some of those were ones I added at about the same time as MMInit, so it
may be that MMInit isn't the culprit for any of the problems I saw.  At the
moment I'm not having problems.
Your mileage may vary.

> ...(remainder of Keith's original message deleted)...

I'm not disagreeing with Keith quite as much as it might sound here.  If you
try the init and it doesn't produce an immediate and unmistakable improvement
in some programs you are running, then there is no reason to use it.  I have
only two programs where the improvement was real (and right now I can't think
what the other program was...), so you're not likely to see much improvement. 
And remember, it only takes one or two system hangs before you'll lose all the
time the init is supposed to be saving you.

But I do think that for me, right now, it's worth it to keep MMInit in my
startup rotation...

 - Garance_Drosehn@mts.rpi.edu
