Date:         Mon, 14 Aug 1995 13:44:01 -0700
Reply-To:     Macintosh Scripting Systems <MACSCRPT@LISTSERV.DARTMOUTH.EDU>
Sender:       Macintosh Scripting Systems <MACSCRPT@LISTSERV.DARTMOUTH.EDU>
From:         "John W. Baxter" <jwbaxter@OLYMPUS.NET>
Subject:      Re: IdleTime INIT crashing 6100/60
Comments: To: Macintosh Scripting Systems <MACSCRPT@caligari.Dartmouth.EDU>

It would be helpful if someone who is using the IdleTime init on a PowerMac
successfully would say so.

It's entirely possible that it won't work (it patches traps, and as far as
I know hasn't been vetted for PowerMac).  On the other hand, there has been
a year and a quarter of opportunity for someone to have reported a problem
before now, and I don't remember any such reports.

I don't run IdleTime (I don't need it, and it patches traps:  therefore I
don't run it).

   --John

--
   It's MY fonts folder, not Adobe's or Microsoft's!
jwbaxter@pt.olympus.net (John W. Baxter)  Port Ludlow, WA

Date:         Mon, 14 Aug 1995 22:28:36 -0700
Reply-To:     Macintosh Scripting Systems <MACSCRPT@LISTSERV.DARTMOUTH.EDU>
Sender:       Macintosh Scripting Systems <MACSCRPT@LISTSERV.DARTMOUTH.EDU>
From:         Doug Baron <dougb@CHILEHEAD.COM>
Subject:      Re: IdleTime INIT crashing 6100/60
Comments: To: Macintosh Scripting Systems <MACSCRPT@caligari.Dartmouth.EDU>

At 1:44 PM 8/14/95, John W. Baxter wrote:
>It would be helpful if someone who is using the IdleTime init on a PowerMac
>successfully would say so.

Full-time.

Doug

Date:         Wed, 16 Aug 1995 12:34:24 -0400
Reply-To:     Macintosh Scripting Systems <MACSCRPT@LISTSERV.DARTMOUTH.EDU>
Sender:       Macintosh Scripting Systems <MACSCRPT@LISTSERV.DARTMOUTH.EDU>
From:         Mike Alexander <Mike.Alexander@UMICH.EDU>
Subject:      Re: IdleTime INIT crashing 6100/60
Comments: To: Macintosh Scripting Systems <MACSCRPT@caligari.Dartmouth.EDU>

>Date:    Mon, 14 Aug 1995 22:28:36 -0700
>From:    Doug Baron <dougb@CHILEHEAD.COM>
>Subject: Re: IdleTime INIT crashing 6100/60
>
>At 1:44 PM 8/14/95, John W. Baxter wrote:
>>It would be helpful if someone who is using the IdleTime init on a PowerMac
>>successfully would say so.
>
>Full-time.
>
>Doug

I've been running it on an 8100 too, and it doesn't patch any traps.  It
installs a Gestalt selector and an jGNEFilter, but no trap patches.


Mike Alexander              Internet:       Mike.Alexander@umich.edu
University of Michigan      America Online: MAlexander
ITD - Research Systems      AppleLink:      UMICH

Date:         Wed, 16 Aug 1995 15:01:39 -0700
Reply-To:     Macintosh Scripting Systems <MACSCRPT@LISTSERV.DARTMOUTH.EDU>
Sender:       Macintosh Scripting Systems <MACSCRPT@LISTSERV.DARTMOUTH.EDU>
From:         "John W. Baxter" <jwbaxter@OLYMPUS.NET>
Subject:      Re: IdleTime INIT crashing 6100/60
Comments: To: Macintosh Scripting Systems <MACSCRPT@caligari.Dartmouth.EDU>

>>Date:    Mon, 14 Aug 1995 22:28:36 -0700
>>From:    Doug Baron <dougb@CHILEHEAD.COM>
>>Subject: Re: IdleTime INIT crashing 6100/60
>>
>>At 1:44 PM 8/14/95, John W. Baxter wrote:
>>>It would be helpful if someone who is using the IdleTime init on a PowerMac
>>>successfully would say so.
>>
>>Full-time.
>>
>>Doug
>
>I've been running it on an 8100 too, and it doesn't patch any traps.  It
>installs a Gestalt selector and an jGNEFilter, but no trap patches.

Thanks, Mike and Doug.  So it would seem likely that some other jGNEFilter
installer on the machine where IdleTime is accused of causing crashes is
conflicting somehow.  [I now forget whose machine that was.]

And it's good to know that IdleTime uses the jGNEFilter:  if I ever *need*
IdleTime, I'll feel better about installing it.

   --John

--
   It's MY fonts folder, not Adobe's or Microsoft's!
jwbaxter@pt.olympus.net (John W. Baxter)  Port Ludlow, WA

Date:         Thu, 17 Aug 1995 03:00:59 -0700
Reply-To:     Macintosh Scripting Systems <MACSCRPT@LISTSERV.DARTMOUTH.EDU>
Sender:       Macintosh Scripting Systems <MACSCRPT@LISTSERV.DARTMOUTH.EDU>
From:         Dave Ely <ely@NETCOM.COM>
Subject:      Re: IdleTime INIT crashing 6100/60
Comments: To: Macintosh Scripting Systems <MACSCRPT@caligari.Dartmouth.EDU>
Comments: cc: "John W. Baxter" <jwbaxter@olympus.net>

John,

 > It would be helpful if someone who is using the IdleTime init on a
 > PowerMac successfully would say so.

 > It's entirely possible that it won't work (it patches traps, and as
 > far as I know hasn't been vetted for PowerMac).  On the other hand,
 > there has been a year and a quarter of opportunity for someone to have
 > reported a problem before now, and I don't remember any such reports.

The IdleTime INIT (v1.0, dated Nov 1993) looks mostly safe, but has a
couple of problems I noticed a while back and which ought to be fixed
(I've just disassembled the INIT I have lying around so current
versions could be different). These could (possibly) be causing part
of the problems people are seeing.

First, it's both a head and tail patch which it doesn't need to be
(all it really needs to do is monitor... let downstream patches do
what they want). The two problems I see are both caused because it
wants to patch both ends... jGNEFilters need to be really defensive
because there are so many which do silly things. It doesn't make sure
it preserves A1 for any patches which are wrapped around it and make
dumb assumptions about being able to look at the event pointer at any
time. Second, it makes a stack copy of the input boolean state before
calling downstream filters, but ignores the return value and pops the
value for D0 from the stack. Since some filters do it right and some
wrong (e.g. this one... the stack is supposed to be passed downstream,
D0 is passed upstream... to 'fix' things you need to know both values
and adjust accordingly). I think it's also pushing and pulling a long,
which might be wrong (should be a short value -- could be wrong
there), around the downstream calls.

Either of these slight idiosyncrasies could cause a problem, although
I would doubt they'd cause crashes all by themselves. The more likely
case is that a combination of this patch with others which are doing
even more bizzare things is causing the crash. Filters are strange
beasts... and here's a strange story to back that up. I had a filter
once which was off by a few bytes (I think it ate 4 bytes -- can't
remember now) on it's return. On any '030 system (mainstream at the
time), it could work for a couple of days before causing a crash. When
run on a 68000 system and it would crash almost immediately. It took
me much longer than it should have before I noticed I had a stack
imbalance which for whatever reason was _not_ causing an immediate
crash on newer systems. These stupid things happen from time to time,
and filters tend to be one place where one's system makes them harder
to find at times.

Anyway, if someone has the sources for IdleTime (Dave, Doug or John?),
please send them my way and I'll take care of the problems I've
mentioned (if you'd like).

 __________________________________________________________________
 Dave Ely                                        ely@metrowerks.com
 Metrowerks, Inc.

Date:         Thu, 17 Aug 1995 06:34:37 -0700
Reply-To:     Macintosh Scripting Systems <MACSCRPT@LISTSERV.DARTMOUTH.EDU>
Sender:       Macintosh Scripting Systems <MACSCRPT@LISTSERV.DARTMOUTH.EDU>
From:         Dave Winer <dwiner@WELL.COM>
Subject:      Re: IdleTime INIT crashing 6100/60
Comments: To: Macintosh Scripting Systems <MACSCRPT@caligari.Dartmouth.EDU>

>Anyway, if someone has the sources for IdleTime (Dave, Doug or John?),
>please send them my way and I'll take care of the problems I've
>mentioned (if you'd like).

That would be Doug. And I think it's great that you're willing to adopt
the IdleTime INIT. It's an incredibly useful thing, I want to include it
with Clay Basket (it actually *is* included with Clay Basket!).

Knowing how long it's been since the user did anything is major stuff.

Dave

Date:         Thu, 17 Aug 1995 11:23:39 -0700
Reply-To:     Macintosh Scripting Systems <MACSCRPT@LISTSERV.DARTMOUTH.EDU>
Sender:       Macintosh Scripting Systems <MACSCRPT@LISTSERV.DARTMOUTH.EDU>
From:         "John W. Baxter" <jwbaxter@OLYMPUS.NET>
Subject:      Re: IdleTime INIT crashing 6100/60
Comments: To: MacScripting@Dartmouth.EDU

>Anyway, if someone has the sources for IdleTime (Dave, Doug or John?),
>please send them my way and I'll take care of the problems I've
>mentioned (if you'd like).

Dave,

   Many, many thanks for taking a look at IdleTime (I've never written one
of those filters, so I'm both innocent of causing problems, and unable to
fix them).  I don't think I have the source (which is much the same as not
having it).

   --John

--
   It's MY fonts folder, not Adobe's or Microsoft's!
jwbaxter@pt.olympus.net (John W. Baxter)  Port Ludlow, WA

Date:         Thu, 17 Aug 1995 15:57:24 -0700
Reply-To:     Macintosh Scripting Systems <MACSCRPT@LISTSERV.DARTMOUTH.EDU>
Sender:       Macintosh Scripting Systems <MACSCRPT@LISTSERV.DARTMOUTH.EDU>
From:         Doug Baron <dougb@CHILEHEAD.COM>
Subject:      Re: IdleTime INIT crashing 6100/60
Comments: To: Macintosh Scripting Systems <MACSCRPT@caligari.Dartmouth.EDU>

At 3:00 AM 8/17/95, Dave Ely wrote:
>The IdleTime INIT (v1.0, dated Nov 1993) looks mostly safe, but has a
>couple of problems I noticed a while back and which ought to be fixed...

Dave,

You're right on here. The INIT isn't actually trying to "tail patch". It
just made the poor choice of JSR'ing to the next filter instead of setting
up the stack correctly and RTS'ing. It doesn't intend to do any further
processing after it let's the next one run.

I'd be happy to send you the source if you have time to make it right.
Watch for it in private email.

Doug