I agree.

All signs indicate that Leopard will be a big release. Tiger has been
out the longest of any OS release (as Apple said when Tiger was put out
in April of 2005, the have slowed down the release cycle to make major
improvements). There are strong indications it will have a vastly
different visual look than Tiger. There are also probably numerous
places where things we currently do don't work correctly or the behavior
changes in the new OS. It's going to mean a bunch more work for us, and
it's work of a scope that is totally unknown, and much of it we may have
to do at the last second.

Another thing I would like to add is that Adium, an alternative IM
client for Mac OS X (that I work on), also has update statistics online
(http://adiumx.com/sparkle). According to those statistics, roughly 5-6%
of users are on 10.3.9. As of this writing, this is roughly the same
number of users as are left on 10.4.8 -- 10.4.9 came out a few weeks ago.

We have been through the decision to drop an OS release with Adium twice
now. There comes a point where the old OS is slowing you down
(development-wise) and preventing you from being competitive. That's
when you know you need to drop the OS. One good litmus? How many of your
community / developer types are using the OS. Right now, neither Josh
nor I have access to a 10.3 machine, despite us trying to load it on our
older PPC PowerBooks (they are too new to load the OS, and they are only
a couple years old). That isn't to say we couldn't *get* such a machine.
The point is that it's fairly old hardware, and those users are probably
not the types to try an alternative browser -- they'll probably just
stick with Safari.

To succeed on the Mac you need to be nimble and fashionable. Josh's work
with native form controls and all the Cocoa port work in general has
gone a long way, and I think with a nice theme and a few "hip" touches,
we can make a big splash with Firefox 3. Dropping 10.3 will help us be
more nimble.

-Colin

Unrelated to the decision to drop 10.3, I think this line of argument is
really really odd.  Maybe that's because of all the people I know who
will buy a computer with some OS, keep using it (with that OS) for a
while (because upgrades tend to break things), but are happy to
experiment with non-OS software, including browsers.  I'm having a hard
time reconciling the statistics being cited with my personal experience,
I guess.  :(

-Boris

That sentence from Colin is the was the only part of the proposal and
accompanying emails I would take issue with, because it is completely
unsubstantiated. It could be correct, but who knows.

I do agree that we shouldn't spend inordinate developer resources on
5-6% of 5-6% of our users. With Linux, we need to be careful because of
the way it overlaps with our developer base (though it doesn't help to
muddle the two).

It's not pleasant to drop support for users that want to use Firefox 3
without paying Apple $100, but engineering is about all about
trade-offs. It doesn't seem like we have the resources to do it all, on
three versions of Mac OS X, and effectively compete with Safari, which
doesn't support older OS X versions for major upgrades.

- Rob, who snidely predicted this day when support for 10.2 was dropped,
and it was promised that 10.3 would be supported for Fx3, but now thinks
we should drop it like a rock

It would be a good idea to see if we (Mozilla) have any stats for what
OS version our OS X users are running. I'm not sure if we collect that
level of detail from downloads/updates, though.

I'd be a little wary of how well the Adium stats correlate with the
actual installed base of OS X.

2 years is a long time in the software industry, but not so much for the
rest of the world. I would assume there is still lots of pre-2005
hardware and software being used in schools and homes... It seems
premature to dismiss it as outdated. [Given that upgrading to 10.4 isn't
free, I'd assume most of that hardware is still running the 10.3 it
shipped with.]

That said, if there are a lot of 10.3 users but they're not showing up
in our FF1/FF2 usage stats, then they're probably not relevant for FF3.

I'm a little wary about this part as well... We're rapidly approaching
the first FF3 beta, which is supposed to be feature complete. Dropping
support for an OS is usually due to core, fundamental issues -- so it
seems rather late in the game to be talking about that.

I can see dropping 10.3 being a win for getting us there faster and
reducing testing requirements, though.

Justin

we should be able to dig this data out.  there are some problems
currently with the loging and reporting system so it might be a few days.

generally the places that report Mac OS version info like
http://marketshare.hitslink.com/report.aspx?qprid=2
don't provide as much detail about OSX versions as they do for windows.

chris h.

Amen.  That's exactly what I see when I look at what non-technical
non-early-adopter users (you recall, the folks we want to expand Firefox
market share with) are using.

If you assume that we're not trying to expand our market share in this
segment, then yes.  Are we making that assumption?

(I'm not saying it's a bad assumption; I just want us to be conscious
that we're making it.)

That's really what the whole thing is about -- lack of resources.  I'm
really pretty unhappy with us dropping OS X 10.3 (because a number of
people I've only recently converted to Firefox are using it), but I can
see why we might have to do it given the date constraints on our Gecko
1.9 release.  :(

-Boris

To clarify, that's with my evangelist hat on.  As a sometimes Mac user
who doesn't use 10.3 himself and wants Firefox to work well on 10.4, I
like the idea, of course.  :(

-Boris

Hooking up my question randomly here:

Given how old the hardware to run 10.3 had to be, how likely is it that
just paying a few 100 bucks is going to make that run 10.4 or .5?

Like, I only know about the hardware requirements evolution on windows
releases, and linux, to some extent. If OSX is only halfway close, we're
not asking users to upgrade their OS, but their hardware.

Axel

On 5/22/07, Colin Barrett <cbarr...@mozilla.com> wrote:

Let's talk about this with more detail, because I'm sympathetic to the
economic argument, especially given our scarce Mac hacking resources,
but I think we need to be more deliberate about it than just pursuit
of fashion and general nimbleness.

Which of the planned Gecko 1.9 features (blockers and otherwise) are
hard to do on 10.3?  How much longer, roughly, would they take to do
in a 10.3-compatible way?  Which parts of the feature would become
infeasible due to missing needed support in the OS?  That information
will help us make a more widely-informed decision about the cost part
of the equation, I think.

Also: what is Apple's own support lifecycle for operating systems?
Will people still be getting security updates for 10.3 when Fx 3
ships, say in Q4 of this year?

Mike

As many of you know, I still run 10.3 on my laptop, mostly to catch PPC
bugs since everyone in the QA team runs Intel Mac with 10.4.x. I don't
buy the argument that if devs aren't using it, it isn't being used by
the community - in fact, Josh and others know I routinely raise issues
on 10.3 just so he doesn't have to run it (Do any of the Windows devs
run Win 98 or 2K - no, but we still support it). I would like to see
some other data to really get a sense of how many users are out there.
In the past I have been asking marketing to see if they can get us some
data on how many 10.3 users we actually have, but I have not gotten
anything.  More data would be useful.

And as shaver says, more information about specific features that would
be hard to do on 10.3 (and are there any workarounds?) as well as
Apple's own support lifecycle would be helpful data to have.

I want to have a kick-ass Firefox 3 that makes Safari quake in its
boots, but I also don't want to dismiss the loyal Firefox users that
still may be using 10.3. We should definitely try to get some of the
data posters have asked about.

Justin Dolske schrieb:

Not only schools and home, but esp. also businesses. Around here, for
example, as a business, you only can deduce taxes for a computer over
its expected lifetime, which is standardized to 4 years by law. Which
means, if a business upgrades its computers faster than every 4 years,
it might more or less lose parts of tax deductions - and 2 years are
only half of that time!

It's one thing to drop Win9x when Win2k debuted in 2000 and XP in 2001,
or to drop GTK1 when GTK2 was released in 2002, but asking users of
computers that are two years old (or only slightly older) to upgrade
their hardware seems a bit drastic. And I'm not sure how much of the
10.3 hardware can still run with Tiger (if Apple even still sells Tiger,
and not just Jaguar).

I guess the real question is how much overhead 10.3 support is actually.

Oh, and BTW, I think our main Mac developer (mostly doing UI stuff only
though) in the SeaMonkey project, Stefan Hermes (stefanh on IRC), still
has 10.3.9 available.

Robert Kaiser

For the record the original Mac Mini (1.42GHz G4) shipped with 10.3
(towards the end of 10.3's life).  So there are still pretty modern PowerPC
based systems running 10.3.  I'm almost positive G5's also shipped with
10.3.  Odds are most of those are still in active use, and likely running
10.3.

Agreed.  Though one should differentiate between hardware and software.
Tiger (10.4) does support my B&W G3 (400MHz) from 1999.  As a general rule
Apple supports hardware for at least 4 years.  Here's a list of hardware
supported (officially) in 10.4:
http://www.apple.com/macosx/upgrade/requirements.html

So upgrading hardware is a bit of a stretch.  Anyone likely running Firefox
has the adequate hardware, they just need an OS update.  For businesses,
this is more problematic because of multiple licenses, IT testing
compatibility, and OS rollouts aren't exactly easy... hence many stall as
long as possible.  

That said, I personally think it's to early to drop 10.3 support.  It
shipped to recently and on relatively modern hardware.  Unless there's a
real lack of resources, or large overhead, I don't think it's a great
choice.

This is the BIG question.

--
Robert Accettura
rob...@accettura.com

On 23-May-07, at 9:30 AM, Robert Kaiser wrote:

We're running around in circles here without any data, and it's not  
really getting us anywhere. I could, for instance, argue from  
anecdotal evidence that OSX users tend to refresh their OS far more  
frequently than Windows users. Seems true, and feels true in my gut,  
so by Colbert's Maxim, it must be right! :)

It's a question of balance, and the right questions have already been  
asked:

  - what is the overhead pain of 10.3 support (in terms of  
development and QA)
  - what is the expected lifecycle of 10.3 as stated by Apple?
  - what is the expected uptake of 10.4 and when is it planned on  
being released?

Right now it feels premature to me to ditch 10.3 support if our  
release of Firefox 3 is planned within the same year as the 10.4  
release, but let's work on getting some answers.

cheers,
mike

Mike Shaver skriver:

For the record: Apple still delivers security updates to 10.3.9 users
(last one was 1 May).

/Stefan

On May 23, 7:08 am, Mike Beltzner <beltz...@mozilla.com> wrote:

well, i think 10.4 has been out since the end of April 2005, and it's
actually 10.5 that's the upcoming release (now slated for november).
10.3 was released in October 2003.

my own anecdotal experience would suggest that folks who bought their
machines in the first half of 2005 even (which would include both my
mother & my mother-in-law) are still running 10.3, as it would have
required some proactive activity on their part to make it different.
(i'm such a lousy son & son-in-law that i haven't gone to upgrade them
yet. holding out for 10.5.)

while far from being actual data, i suggest that there are an awful
lot of folks who never shelled out for 10.4.

On 23-May-07, at 6:27 AM, Mike Shaver wrote:

I'm going to get even more aggressive on the subject:

Do we want to support a platform that the vendor has dropped security  
updates for?  I would assert that we do not.  We've historically  
dropped platforms as they hit end of life.
Do we want to support platforms that are about to be dropped shortly  
before/after targeted release dates?  I would assert that we don't  
want to do this either (Firefox 3, if released in November, will be  
supported until May 2009, so thinking only about ship date is a  
little short-sighted).

 From an IRC conversation with Josh, it seems fairly likely that  
Apple will drop support for OS X 10.3 at the end of this year.  Given  
that Firefox 2 will continue to be supported until at least six  
months after Firefox 3 ships, I think we will have the 10.3 holdouts  
covered longer than Apple, and I think that's all we should  
realistically plan to do.

-- Mike

I asked some friends who work for a Mac repair/reseller shop, and they
follow this sort of stuff rather closely.  G4s will in general run
10.4.0 no problem, however Apple put in a hard shut-off even for some of
these in the 10.4.7-ish time frame, so OSes newer than that would not
work on some G4 hardware.  The G3 will not run 10.4 at all without
modding, and since laptops ran about a year behind the desktops, some
people do still have G3 laptops that are functional machines for them.
Most people are on G4 or newer, though if they are still on a G4 or G3
with 10.3, chances are they aren't going to install a new OS: they're
waiting for a hardware upgrade to do so.  Asking them to upgrade their
OS really is asking them to upgrade their hardware.

On 23-May-07, at 11:32 AM, John wrote:

Whoops - yeah, sorry, I got my numbers wrong there ;)

cheers,
mike

On May 23, 5:27 am, "Mike Shaver" <mike.sha...@gmail.com> wrote:

At least most of that information is in my original proposal.

That is definitely in my original proposal. Colin's post is a response
to that, please read the original.

We're making this assumption that Apple will drop support. It's a fair
assumption, to be sure, but it's an assumption nonetheless. Has Apple
officially said that they'll be dropping support?

-Sam

Out of curiosity, when did they drop 10.2 support?  I just tried to find
that information and failed...

-Boris

Apple is almost certainly not going to deliver security updates to
10.3 users after this year. They'll stop around the time we ship Gecko
1.9. I can't find an official position on that but I don't recall
finding one on 10.2 either.

Secondly, on the hardware/software upgrade issue, users don't *have*
to run Firefox 3 just like they don't *have* to run Mac OS X 10.4. If
you really want Firefox 3 for some compelling reason you can get it -
install 10.4. But there are other good browsers for 10.3, like Firefox
2 which we're all so fond of. Your mother in laws wouldn't be getting
left in the cold by us any more than they would be by Apple.

Josh asked me to weigh in on this, since it will have a significant
impact on Camino development.

On May 23, 8:32 am, John <john.li...@gmail.com> wrote:

There's no question that there are people that are running 10.3 who
cannot or will not upgrade (heck, Camino's feedback list gets requests
for an OS 9 version every month or two), and I don't think there's any
question that we would like to continue to support them in an ideal
world. We've always made a real effort to support older OS versions
for Camino for as long as we reasonably can. All other things being
equal, we'd much rather have the option of supporting 10.3 for Camino
2.0 (when we will be picking up Gecko 1.9).

However, all other things aren't equal; Gecko 1.9 is a huge overhaul
of major parts of the tree. 10.3 has already caused problems in the
printing rewrite, in places touched by Cairo changes, and now in the
widget rewrite that will substantially improve the feel of 1.9 on OS
X. There are still substantial bugs that need to be resolved to make
1.9 solid, and I'm in agreement with Josh that given the limited
resources, spending time on the 10.3-only bugs that have resulted is
probably not realistic at this point.

If it were just a case of "let's drop 10.3 to make our lives a little
easier" I wouldn't be for it, but looking at the bug lists the
question I see is "is it worth sacrificing overall quality of 1.9 to
continue to support 10.3", and I think the answer there should be
"no", as unfortunate as that is.

-Stuart

Josh asked me to weigh in on this, since it will have a significant
impact on Camino development.

On May 23, 8:32 am, John <john.li...@gmail.com> wrote:

There's no question that there are people that are running 10.3 who
cannot or will not upgrade (heck, Camino's feedback list gets requests
for an OS 9 version every month or two), and I don't think there's any
question that we would like to continue to support them in an ideal
world. We've always made a real effort to support older OS versions
for Camino for as long as we reasonably can. All other things being
equal, we'd much rather have the option of supporting 10.3 for Camino
2.0 (when we will be picking up Gecko 1.9).

However, all other things aren't equal; Gecko 1.9 is a huge overhaul
of major parts of the tree. 10.3 has already caused problems in the
printing rewrite, in places touched by Cairo changes, and now in the
widget rewrite that will substantially improve the feel of 1.9 on OS
X. There are still substantial bugs that need to be resolved to make
1.9 solid, and I'm in agreement with Josh that given the limited
resources, spending time on the 10.3-only bugs that have resulted is
probably not realistic at this point.

If it were just a case of "let's drop 10.3 to make our lives a little
easier" I wouldn't be for it, but looking at the bug lists the
question I see is "is it worth sacrificing overall quality of 1.9 to
continue to support 10.3", and I think the answer there should be
"no", as unfortunate as that is.

-Stuart

On May 23, 9:08 am, josh...@gmail.com wrote:

yep, totally right. fair enough. but our basic plan is to do security
updates for fx2 for 6 months past the release of fx3, right? so that
puts us into middle of 2008, or about 3 years after 10.4 was released.
that's probably okay.