93195 - We shouldn't be blocking Mac OS 8.5 [need to fix comment]

Status: RESOLVED FIXED

(SeaMonkey :: Build Config, defect, P3)

Description

[Asa Dotzler [:asa]]

I was all ready to do a Mozilla 0.9.3 release when I discovered that Mozilla
will not start on Mac OS 8.5. It worked just dandy on Mozilla 0.9.2 and now it
fails with an error that says I must have 8.6 or better.  I couldn't find a
Bugzilla bug or newsgroup discussion about this and I don't understand why it
would work just fine 5 weeks ago and now it's completely blocked.  I hope this
is an easily corrected mistake.

I'm holding the 0.9.3 Milestone release until I understand better what happened
here.

Comment 1

[cls]

According to lxr, pchen checked in that change:
http://bonsai.mozilla.org/cvsblame.cgi?file=mozilla/xpfe/bootstrap/nsNativeAppSupportMac.cpp#182

Comment 2

[Steve Dagley]

Marking WONTFIX.  It's blocked because Mac OS 8.6 is supposed to be the minimum 
supported version of Mac OS for Mozilla.  This was decided amongst the Mac 
community at Netscape some time ago (January IIRC) for a few reasons:  

- ATSUI, which we don't rely on for text rendering but will use in some 
circumstances, is buggy in 8.5.

- CarbonLib 1.1 and later require Mac OS 8.6 or later.  Note this was an issue 
when we were going to have a single Carbonated binary for Mac OS 8/9 but 
currently we're not going that route.

- Simplifies the test matrix.

- The upgrade from 8.5 to 8.6 is free.  If someone isn't going upgrade their OS 
do you expect them to upgrade their browser?  (To paraphrase the response I got 
from Apple when I asked why networking fixed made in 9.x weren't made available 
for people running 8.x)

   

Comment 3

[Patrick C. Beard]

And yet, Mozilla 0.9.2 has been observed to work just fine on 8.5. We could back
off and say it is unsuppported on OS's earlier than 8.6, and perhaps warn users,
but if it in fact runs, why not let it? I think at least in Mozilla builds, we
should be more permissive.

Comment 4

[Mike Pinkerton (not reading bugmail)]

it's been this way for about a year, at least and people are just _now_ noticing? 
Seems like it's not a big deal to require 8.6 then.

Comment 5

[Mike Pinkerton (not reading bugmail)]

oops, i didn't realize paul didn't change the check until recently. we've been 
saying 8.6 for at least a year IIRC. My bad.

Comment 6

[Daniel (Leaf) Nunes]

we're backing it out at least on the branch (the check for 8.6, that is). can i 
get a review?

Index: mozilla/xpfe/bootstrap/nsNativeAppSupportMac.cpp
===================================================================
RCS file: /cvsroot/mozilla/xpfe/bootstrap/nsNativeAppSupportMac.cpp,v
retrieving revision 1.10
diff -u -r1.10 nsNativeAppSupportMac.cpp
--- nsNativeAppSupportMac.cpp   2001/07/13 22:40:18     1.10
+++ nsNativeAppSupportMac.cpp   2001/08/02 17:35:03
@@ -181,8 +181,8 @@
 {
   long response = 0;
   OSErr err = ::Gestalt (gestaltSystemVersion, &response);
-  // check for at least MacOS 8.6
-  if ( err || response < 0x860)
+  // check for at least MacOS 8.5
+  if ( err || response < 0x850)
   {
     ::StopAlert (5000, NULL);
     return PR_FALSE;

Comment 7

[Asa Dotzler [:asa]]

*- ATSUI, which we don't rely on for text rendering but will use in some 
circumstances, is buggy in 8.5.*

I don't know what this means. I have been testing 0.9.2 (and previous Milestones
back to 0.7) on Mac 8.5 and don't encounter any problems in my normal usage
except problems that also affect 9.1.  What bugs do we have on file that totally
break 8.5? 

*- CarbonLib 1.1 and later require Mac OS 8.6 or later.  Note this was an issue 
when we were going to have a single Carbonated binary for Mac OS 8/9 but 
currently we're not going that route.*

So this is no longer a justification?

*- Simplifies the test matrix.*

Netscape's QA resource shortage is not justification for checking in code which
has no purpose except to break Mozilla for a particular operating system,
especially a system on which Mozilla has been working without major problems for
all previous Milestones. Netscape Client QA doesn't support a lot of the OSes
that Mozilla runs on but we don't intentionally break Mozilla for those systems.
If Netscape decided that they didn't want to test or support RedHat 6.1 would
the community tollerate someone checking in code which served no purpose except
to break Mozilla that has always worked fine on that OS? We have thousands of
bug reports against 8.5  There are obviously 8.5 users testing our Milestones.

*- The upgrade from 8.5 to 8.6 is free.  If someone isn't going upgrade their OS 
do you expect them to upgrade their browser?  (To paraphrase the response I got 
from Apple when I asked why networking fixed made in 9.x weren't made available 
for people running 8.x)*

The upgrade to RedHat 7.0 and 7.1 is free and I can tell you that there are tens
of thousands of RH6.1 users downloading Mozilla Milestones and reporting bugs
which affect other versions of Linux as well as win32, Mac, BeOS, etc. 
Intentionally driving away people who have been downloading, testing and
reporting bugs against our builds is wrong.  Where was this discussion? Why
wasn't I a part of it? Where are the bug reports about Mozilla being so broken
on 8.5 that we had to finish it off with this checkin? Where is the bug report
for checkin in this "unsupport 8.5" fix? 

Removing it from the list of operating systems that Netscape supports is a
decision that Netscape is free to make but that code belongs in the Netscape
tree and not Mozilla. Until such a time as Mozilla no longer works on 8.5 and no
one steps up to fix it I think it's wrong to intentionally break it.

Comment 8

[Blake Ross]

sr=blake for the branch

Comment 9

[Claudius Gayle]

Somebody  should ask NetscapeQA about NetscapeQA :-) <grumble>I have been
running mozilla on MacOS 8.5.x the whole damn time. Heck, I ran 8.1 until you
took that away too. All of my tests, testing, and analysis have always and will
continue to support and be 8.5.1 compliant. I'm not the only one. We only
noticed today because NetscapeQA just went back to the trunk and this checkin
must've happened while we were all looking at the branch.</grumble>

Comment 10

[Claudius Gayle]

Are we gonna do this on the trunk? Are we gonna do it soon? i have builds that
don't run...

Comment 11

[Asa Dotzler [:asa]]

If the only question here is how we're going to do this so that Netscape can
block 8.5 users then lets check this into the trunk so people can get back to
testing and reporting problems (that probably affect 8.6, 9.1, and other
platforms too) and let someone back it out when Netscape has does a branch.
Anyone object?

Comment 12

[Paul Chen]

Attachment: change 0x860 back to 0x850

Comment 13

[Paul Chen]

I've also fixed ErrorAlerts.rsrc but I fear attaching a binary file into
mozilla, ESPECIALLY a mac file with a resource fork.

Awaiting r= and sr= for trunk landing

Comment 14

[Kathleen :Brade]

r=brade

Comment 15

[Paul Chen]

marking p1

Comment 16

[Simon Fraser [no longer active]]

sr=sfraser

Comment 17

[Paul Chen]

fix checked in

Comment 18

[Jens-Uwe]

please look at
http://lxr.mozilla.org/seamonkey/source/xpfe/bootstrap/nsNativeAppSupportMac.cpp#184

the comment says, that we do a check for MacOS 8.6, while we check for MacOS8.5.
I think this is wrong, the comment should match the code.  The patch by Daniel
Nunes did it right...

Comment 19

[Paul Chen]

marking mozilla0.9.5 and minor

Comment 20

[Jatin Billimoria]

Is 8.5 going to be supported in the 0.9.4 branch builds?

Comment 21

[Peter Trudelle]

Mass-moving lower-priority 0.9.5 bugs off to 0.9.6 to make way for remaining
0.9.4/eMojo bugs, and MachV planning, performance and feature work. If you
disagree with any of these targets, please let me know.

Comment 22

[Kathleen :Brade]

adjusting summary to match current bug issue

Comment 23

[Claudius Gayle]

from reading the bug I can't exactly sort out what happened here. Let me say
though that I have a 2001092011 branch build and it can't run on MacOS 8.5.1. I
get a little dinky dialog telling me that I need at least MacOS 8.5 to run.

I guess we're still checking for 8.6 on the branch? Didn't we decide somehwere
we didn't have to?

Comment 24

[Peter Trudelle]

Claudius, do you really think this is worth changing the branch for at this
point?  Is everyone confident that our 8.5 user(s) would be happy with the
product?  Anyone know how many folks besides Claudius are using 8.5x?

Comment 25

[Asa Dotzler [:asa]]

Netscape users on 8.5 would probably be fine with Netscape 6.next.  I test most
of my non OS X bugs on 8.5 and any bugs I see reported against 9.1 existon 8.5
and any on 8.5 are also on 9.1.  Everyone in the world except me and Claudius
and a couple of other Mozilla testers has upgraded to something newer so
Netscape probably does't need to worry about it. 

Comment 26

[Peter Trudelle]

Agreed, nsbranch-, lets just make sure this works right on trunk.

Comment 27

[Paul Chen]

If you look at the comment by Jens-Uwe on 2001-08-23, this bug was reopened
because I forgot to change the *comment* back to 8.5 on the trunk. As far as I
know, the trunk will still run under macos 8.5. Marking p3, minor, and mozilla0.9.6

Comment 28

[Paul Chen]

Attachment: fix comment to match code

Comment 29

[Paul Chen]

Comment on attachment 53936 [details] [diff] [review]
fix comment to match code

verbal r=jag and rs=ben

Comment 30

[Paul Chen]

comment fix checked in