> it's about how much pain it is to support 10.4, 10.5, 10.6 32bit
> and 10.6 64bit all at the same time,
I think you should also take in account the Intel/PPC support.
It means that keeping 10.4, Mozilla have to test/support: 10.4 PPC 10.4 x86 10.5 PPC 10.5 x86 10.5 x64 10.6 x86 10.6 x64
Abandoning 10.4, means 29% of OS/CPU combination less.
If concerns arise about the 10.4 user base, extend the security-fix support of 3.6 from mid-2011 to end-2011, it should be less work (but still work) to maintain an old branch for security rather than to back-port anything new; especially as other products (Tb?) may still need Gecko 1.9.3 by then.
Note that I don't know if you plan to support x64 on 10.5. I don't think it is useful.
On Feb 4, 9:16 pm, Josh Aas <josh...@gmail.com> wrote:
> Where > we can work around supporting 10.4, doing so consumes valuable time > and effort. Neither Chrome nor Safari has to deal with this.
Actually this is factually incorrect for Safari: the current version of Safari is 4.0.4. It is fully supported on OS 10.4.11 on PPC and Intel Macs. I am running it right now on a 1999 G4 AGP 450 MHz, which is my primary OS X Mac. I support a number of clients/family/friends, few of whom have Intel Macs and most of whom are running Tiger (and this same version of Safari). Safari V.4 is actually working better on the old Macs i deal with (directly or indirectly) than late versions of Safari 3.
Still, that number of users is invisible compared to the numbers of users of other OS versions/platforms that the people working on Mozilla are dealing with. While it would be nice to see continued support for Tiger and while i am a fierce trailing-edge user who champions keeping older systems going as long as possible, i am also well aware that legacy support can be at least a drain and possibly a project killer as has been pointed out. I personally trust those who are active with the Mozilla project to make wise decisions to keep a reasonable balance, so while i vote very strongly for Tiger support and would personally benefit from it (as would my clients/family/ friends nearly to a person), i have no hard facts/numbers to sway the discussion and do understand that it may be in the best interests of the project and majority of users to drop Tiger and move forward.
> On Feb 4, 9:16 pm, Josh Aas<josh...@gmail.com> wrote: >> Where >> we can work around supporting 10.4, doing so consumes valuable time >> and effort. Neither Chrome nor Safari has to deal with this.
> Actually this is factually incorrect for Safari: the current version > of Safari is 4.0.4. It is fully supported on OS 10.4.11 on PPC and > Intel Macs.
You mixed it. The discussion is not going about Firefox 3.6.x supporting OS 10.4 or not (it does, and will), but about Firefox.next (being 3.7 or 4.0), supporting it.
The comparison to Safari lead then to the question: "Will Safari *5* support 10.4?"
I don't have the answer (Apple doesn't comment on future products).
But my sentiment is no, Safari 5 will very likely not work on OS 10.4, neither on PPC. It may even be a 64-bit only release.
> You mixed it. The discussion is not going about Firefox 3.6.x supporting OS 10.4 or not (it does, and will), but about Firefox.next (being 3.7 or 4.0), supporting it.
> The comparison to Safari lead then to the question: > "Will Safari *5* support 10.4?"
> I don't have the answer (Apple doesn't comment on future products).
> But my sentiment is no, Safari 5 will very likely not work on OS 10.4, > neither on PPC. It may even be a 64-bit only release.
Fwiw, the latest nightly builds of WebKit still run (fine) on 10.4.11 (I can't test how well they run, but they do run).
> At the time, it was ~3.5 years since 10.3 had been released, Firefox 3.0 > was expected to be ~6 months away (reality was 1 year). AUS data > indicated that 16% of Firefox users were running 10.3.
> Now it's ~3.5 years since 10.4 was released, ?? months until Firefox ?.? > ships, and Josh's data indicates that 25% of our current users are on > 10.4 (but that early signs are that it's much lower for 3.6).
> So, offhand it seems reasonable to me, and roughly comparable to what we > did w.r.t 3.0/10.3. ><snip>
I'm confused. AIUI, per what Boris said, ?.? (from trunk) will ship in Q2/Q3, which is at most 2/3 of a year away. 25% of users rather than 16% of users use it. If I had to guess at the overall size of the mac Fx userbase, I'd say it has grown since that 3.0 decision point (if someone has solid data from AUS, please share).
So, all in all, if we drop 10.4 support, we remove support for a significantly greater (10 pct points) minority of our userbase, which is also a significantly larger group of people (in absolute terms), at a significantly (33%) faster pace than when 10.3 support was removed.
I don't see how that is "roughly comparable". What am I missing?
I'm not saying it's not the right thing to do because I'm honestly not sure what the cost of the 10.4 support is. I do think the numbers are food for thought.
~ Gijs
(for full disclosure: I'm typing this on a 10.6 macbook, which was upgraded from 10.4 using the $30 upgrade dvd your average apple store will sell you. Obviously won't work for 10.4 ppc users - would be interesting to see how many of them there are?)
> I would object, since I am using an older PowerPC PowerBook under > 10.4, and even 3.6 seems a bit slow or buggy to me.
This seems to indicate that we can't effectively support 10.4 already, doesn't it? So are you suggesting that we should /increase/ the investment in 10.4 support?
<gijskruitbo...@gmail.com> wrote: > I'm confused. AIUI, per what Boris said, ?.? (from trunk) will ship in > Q2/Q3, which is at most 2/3 of a year away. 25% of users rather than 16% of > users use it. If I had to guess at the overall size of the mac Fx userbase, > I'd say it has grown since that 3.0 decision point (if someone has solid > data from AUS, please share).
10.4 users would still have a supported release until FF 3.6 was end-of-lifed, which I would expect to be at least 6 months after the trunk release of which Boris speaks. They wouldn't be able to upgrade to the latest and greatest, but they would still get stability and security updates.
On Sunday 2010-02-07 00:19 +1300, Blair McBride wrote:
> There may well be 1 million users using 3.5 and 3.6 in a year's > time. And those versions will continue to support 10.4. But the > majority of those on 10.4 won't upgrade to 3.next, even if it has > 10.4 support. As it is, very few 10.4 users upgraded to 3.6. Looking > at the numbers, 24% of 3.5 Mac users run 10.4; while only 12% of 3.6 > Mac users run 10.4. I'd say 6% on 3.next would be a very optimistic > number - and at great development cost.
I think numbers for Firefox 3.6 users will change substantially once we make a prompted major update offer from 3.5 to 3.6. Please note that the 3.6 numbers Josh posted were much smaller than the 3.5 numbers (basically 5.2% of the 3.5+3.6 users were on 3.6, and 94.8% were on 3.5).
The 3.6 users right now are presumably largely people who are interested enough in technology that they read sources of news that told them about the 3.6 release, users who I'd expect are more likely to be on the cutting edge in terms of OS versions as well.
L. David Baron wrote: > On Sunday 2010-02-07 00:19 +1300, Blair McBride wrote: >> There may well be 1 million users using 3.5 and 3.6 in a year's >> time. And those versions will continue to support 10.4. But the >> majority of those on 10.4 won't upgrade to 3.next, even if it has >> 10.4 support. As it is, very few 10.4 users upgraded to 3.6. Looking >> at the numbers, 24% of 3.5 Mac users run 10.4; while only 12% of 3.6 >> Mac users run 10.4. I'd say 6% on 3.next would be a very optimistic >> number - and at great development cost.
> I think numbers for Firefox 3.6 users will change substantially once > we make a prompted major update offer from 3.5 to 3.6. Please note > that the 3.6 numbers Josh posted were much smaller than the 3.5 > numbers (basically 5.2% of the 3.5+3.6 users were on 3.6, and 94.8% > were on 3.5).
> The 3.6 users right now are presumably largely people who are > interested enough in technology that they read sources of news that > told them about the 3.6 release, users who I'd expect are more > likely to be on the cutting edge in terms of OS versions as well.
> -David
I am type person that updates either automatically through auto update or going to web site and updating. In in the Past I tested nightly until I figured what I said about bugs or suggestion were laughed at or ignored. since users don't know anything.
> If you currently have limited resources (programmers), try using some > of the $25 million or more you collect every year from Google to hire > more Mac OS X programmers. Not only would that add to the expertise > you have, it would make your development cycle move faster.
Have you read The Mythical Man Month? Adding more people will not make us move faster in the short term, and it may not in the long term either.
Besides that, you are arguing that we should spend a large sum of money on small percentage of our users (a shrinking percentage at that). What would you say to all of our windows users who would then see less resources devoted to them?
> You might not currently have the resources you need to support all > three versions, but you have more than enough money to get those > resources; all it takes is the courage to do it and the willingness to > support your users. Alienating users by dropping support when there's > no valid technological reason only makes more users unhappy and causes > them to switch again to another browser, which is a never-ending > cycle.
I'm sorry, but Josh covered the technical reasons in his first post. You can certainly try to argue that they aren't issues, but simply dismissing them is not the way to do that.
> I can't see the future, so I have no idea what the usage numbers will > be, just as you have no idea. I'm sure you're well aware that Apple > doesn't comment on future plans and I certainly can't make them talk. > I do know that there are millions of Mac OS X 10.4 users who either > cannot or will not upgrade, for a variety of reasons, all of them > perfectly valid for their needs. I also know that Mac OS X 10.5 and > 10.6 are riddled with serious bugs and had features removed without > notice; something Apple doesn't publicize or disclose to anyone before > or after they purchase the OS or a system with those versions > installed. Users are left to discover it on their own, and it's always > when they need it to work correctly; that's too late.
What are these serious bugs and removed features you speak of? If you are trying to convince us that this is a serious issue, I suggest you make actual claims instead of hand-wavy references to problems that come across as FUD.
> So the question you need to ask yourselves before pursuing this ill- > considered idea is: "Do you want to spend the money to hire the > programmers you need to support 10.4, or do you want to give up > millions in revenue?"
It's not even about hiring programmers. It's also about buying more hardware, and supporting it. It's also about using our resources in the most beneficial way for our users. You are arguing that we should spend a disproportionate amount of our resources on a small and shrinking (even if the number of 10.4 users stays the same, they will make up a smaller percentage of our users over time) percentage of our user base. We are not a massive corporation like Apple, so we have to use our limited resources in the best possible way for our users.
> I don't see how that is "roughly comparable". What am I missing?
I suspect the 25% figure is high, because there's not much 3.6 data yet. The 3.6 data currently has only 12% of OS X users on 10.4... No great surprise that early adopters are not running 10.4. As 3.6 gains uptake, I'd expect 10.4 percentages to level off somewhere less than 25%, and of that a similar fraction of users wouldn't be upgrading to 3.7/4.0 anyway.
I say "roughly" comparable because, well, they seem like roughly similar percentages! :) I don't think these kinds of decisions require very precise comparisons. "About a fifth" seems close enough to me.
JimR63701 wrote: > Robert, your own bias appears to be showing... (Not meant as a > criticism, but the reality is, it's not "your ox that's being gored", > as you don't seem to have any personal investment on behalf of the Mac > version)
It might, and I'm neither on a Mac nor on an even more-used OS in terms of our numbers. Hell, I'm not even a Firefox user, but still a Gecko-lover and Mozilla enthusiast. In my project, 1.4 million daily users would be more than our application has as a whole. But in Firefox world, 1.4 million daily users on OS X 10.4 are not only less than a quarter of all Mac users, but also an almost (!) insignificant number compared to the total daily user number that is somewhere between 200 and 300 million, IIRC. Of course, it's not completely insignificant or 10.4 would not work with any current versions. It's even important enough that the just-released Firefox 3.6 will - for its whole lifetime of probably still at least a year - support this OS version fully, with add-ons, security updates and everything. The same will probably be true for Thunderbird 3.1.
All the talk here is about the *next* version of Firefox and other Mozilla applications, i.e. Firefox 3.7 (or higher), Thunderbird 3.2 (or higher) and probably SeaMonkey 2.1 as well, all of which will not be released before summer or fall this year, maybe even later. Note that all future version numbers are tentative and can be changed at any time (if so, usually to a higher number).
> If whatever is shipping off trunk ships later than about end of Q3 or > beginning of Q4 of this year, then imho we've screwed up.
I know the team insist it will be able to meet that schedule, but it only insisted till the very end that it would be able to meet the November 2009 schedule for 3.6
IMO you will start stabilizing 3.7 toward the beginning of Q4, and release it very end of 2010/early 2011. At best. But there's nothing horrible in that. That's about the minimal interval between versions that users can handle comfortably.
> I know the team insist it will be able to meet that schedule
I think you have this backwards. My claim is that we should scope work so as to fit that schedule.
Note that there is no "the team" here; there is just me making a claim about what I think we should aim for.
> but it only insisted till the very end that it would be able to meet the > November 2009 schedule for 3.6
I'm not sure what you're lumping under "the team" here, but the issue there was partly picking the scope first and partly not stabilizing until too late.
> IMO you will start stabilizing 3.7 toward the beginning of Q4, and > release it very end of 2010/early 2011.
If we start stabilizing at the beginning of Q4 (so that would be October 2010) then we are not likely to ship anything until April 2011; more likely May 2011.
We need to start stabilizing in May 2010 at the latest, again imo.
> But there's nothing horrible in that.
There is in terms of PR and being perceived as falling behind in standards support and performance.
> That's about the minimal interval between versions > that users can handle comfortably.
So you think the interval between 3.5 and 3.6 was too short to handle comfortably?
> Nonsense. On the one hand you're arguing that you don't have the > resources, and on the other you're saying that if you did have those > resources it wouldn't help you.
Wouldn't help in the timeframe involved here.
That is, if we hired more people _today_ they would not be up to speed in time to help.
In the long term, the main issue is that amount of work that can be done scales sublinearly with number of people doing it. So in fact putting more people on an issue requires "disproportionate" investment to get an improvement, after a certain point.
> I'm not arguing that you should spend a disproportionate amount of > resources on supporting 10.4 at all. I'm arguing that you already have > the hardware to support Mac OS X 10.4
So you're arguing that support for 64-bit builds on 10.6, say, is less important than support for 10.4? The point is that the resources currently being used for 10.4 would be repurposed for better supporting 10.5 and 10.6. In particular, the hardware used for 10.4 and the hardware needed for 64-bit 10.6 builds are about equivalent in terms of number of machines and such.
> In the meantime, you're trying to convince people that you're > suddenly going to lose those resources after 3.7 ships.
We are if we want to use them to support 10.6 64-bit builds.
> you just don't want to use it to support 10.4 users
Right, because we feel that there is a bigger user demographic that would be better served by it.
Note that the hardware is not a huge issue, by the way; the fact that you have to structure the code very differently to work on 10.4 and 10.6 (_especially_ 10.6 64-bit) is a much bigger problem.
> You could just as > easily drop support for Firefox for Windows and see if that forces all > those users to get a new Mac. I'm betting it won't.
Of course it won't. I don't see what that has to do with the discussion here, though.
> The notion that your 10.4 users are a small percentage of Firefox > users may be true, but at this point you can only infer that user base > will shrink
That seems like a good bet, since the actual number of 10.4 users is likely to be shrinking. It's not like people are buying many new 10.4 installs out there.
> You don't have billions in cash like Apple, but you do have a lot more > money than indie software developers, so you should make your product > work on as many versions of Mac OS X as possible
That's a non-sequitur, sorry. We should make our product work as well as it can for as many of our users as it can. The question is what to do when those two goals are in conflict, as here. We can significantly improve the user experience on 10.5 and especially 10.6 if we drop support for 10.4 (we're talking something like 30% performance improvement on 10.6, for example if I recall the numbers correctly, between the newer compiler and doing 64-bit builds). Which of those is more important? It really depends on one's point of view, obviously.
> especially if Apple > is still supporting those versions with security updates or browser > updates, as they are with 10.4.
No one's talking about dropping 10.4 support right now. The discussion is about dropping support for it in late spring 2011 or so at the earliest.
Sadly, that involves guessing what Apple will do. Guessing wrong one way means we drop support for users whose OS is still supported. Guessing wrong the other way means we invest a lot of time into users whose OS is no longer supported and shortchange other users.
Welcome to life. It's full of hard choices. What Josh is asking for here is any information anyone might have that has not been brought up yet (of which I have seen none in this thread so far) that will affect the decision that has been made thus far based on the information that has already been raised before.
> You're probably spending more time and resources on supporting a crappy OS > (Windoze), because Microsoft doesn't know how to make a decent OS
And just for the record, we're putting in a lot more work into our Mac support than Windows support, on a per-user basis. If we hired 2 more people to do Mac stuff, we'd be putting in more work in absolute terms too.
The fact is, the changes between Mac OS versions are _huge_; supporting multiple versions of Mac OS at once is a huge pain (supporting 10.4 and 10.6 at once is about like supporting Win98 and Win7 at once, if not harder).
> In in the Past I tested nightly until I figured what I said about > bugs or suggestion were laughed at or ignored. since users don't > know anything.
I'm sorry you had that experience -- no one should be laughing at anyone trying to give good feedback.
> Then you're acknowledging publicly that it doesn't matter what users > say, you're going to do something that you already decided to do > before asking for input. Choices may be hard, but when you ask for > user input, you should at least listen to the wishes of the users, not > your own predetermined decision. Welcome to the ranks of hypocrisy.
There is a difference between having your wishes and opinions listened to and having them change the proposed decision, please don't confuse the two.
The fact of the matter is that the people who originally posted asking for feedback have already researched the pros and cons of this decision to the best of their ability. It is the realm of possibility that they missed some crucial fact, but thus far such an oversight has not been pointed out (as far as I can tell anyway).
> On 2/6/10 11:16 AM, Phillip Jones wrote: >> In in the Past I tested nightly until I figured what I said about >> bugs or suggestion were laughed at or ignored. since users don't >> know anything.
> I'm sorry you had that experience -- no one should be laughing at > anyone trying to give good feedback.
But it is common nevertheless. And 'ignored' is even more common.
johnjbarton <johnjbar...@johnjbarton.com> wrote: > On 2/7/2010 9:20 AM, Daniel Veditz wrote: > > On 2/6/10 11:16 AM, Phillip Jones wrote: > >> In in the Past I tested nightly until I figured what I said about > >> bugs or suggestion were laughed at or ignored. since users don't > >> know anything.
> > I'm sorry you had that experience -- no one should be laughing at > > anyone trying to give good feedback.
> But it is common nevertheless. And 'ignored' is even more common.
Yup. And it only takes one bad bugzilla (or support forum) experience to put a user off contacting us *forever*, & they will tell all their friends not to bother trying to give us feedback, too.
IMO a good start toward fixing this would be to change the Bugzilla etiquette guidelines so that under no circumstances short of actual spam is it acceptable to tell anyone to stop commenting on a bug.
> I find that hard to believe. I know many Mac programmers with years of > experience, having worked for Adobe, Apple, etc on major applications, > include Adobe Illustrator and iMovie. They're out there and available > for the right price.
<shrug>. All I know is that finding competent people who are also willing to work in our environment (e.g. actually working with the community, not getting commit access the day that they get hired, etc) is not being particularly easy. I can't speak to the pricing, since I'm not involved in those decisions.
> Yes, they are less important. 64-bit is strictly hype, there is no > benefit to the average user.
Given that trivial performance measurements show that 64-bit builds are a good bit faster than 32-bit once on just about any metric you care to measure (Sunspider, Dromaeo, pageload performance), I wonder where you're getting your data.
> You don't need to support 64-bit builds, because users gain nothing by > using 64-bit applications.
This is particularly false on Mac OS 10.6, which comes with most of its default apps 64-bit by default. As a result we get the very specific gain that starting Firefox doesn't have to load all the 32-bit compatibility libraries (which significantly improves the cold-start-after-computer-on performance of Firefox), in addition to the general "it's got more registers so the compiler can do a better job" speedups linked to above. Loading those libraries involves a lot more disk access and significantly slows down startup.
> The only exception is a very narrow subset > of all users that use resource-intensive applications such as those > for motion graphics, RAW image processing, and even film editing, that > are designed to get a performance boost from it.
Citation needed.
> It has everything to do with it, because what Mozilla plans to do is > based on the assumption that it will spur 10.4 users to switch to > 10.6
No, there is no such assumption. The only assumption in this vein is that users will either switch to 10.5 or 10.6 or not, completely independently of what we do, and that the net effect is that about 18 months from now the fraction of our Mac userbase on 10.4 will be smaller than either that on 10.5 or on 10.6.
The former assumption seems like a good one to me (I doubt many current 10.5 or 10.6 users will be going back to 10.4). The latter is already true; see the numbers Josh cited. How _much_ smaller the 10.4 user base will is open to debate; predicting the future is hard.
> So if you think that tactic will work with > Mac users, there's no reason you shouldn't do the inverse and impose > it on Windoze users and see if they buy a Mac instead.
This suggestion seems to be based on a fundamental misunderstanding, per above.
> Then you're acknowledging publicly that it doesn't matter what users > say, you're going to do something that you already decided to do > before asking for input.
No. I'm saying that we're looking for _new_ input, that hasn't already been brought up the last time this conversation happened and hasn't already been taken into account.
> Choices may be hard, but when you ask for > user input
Josh asked for "input". That's not necessarily the same thing as "user input"; he's looking for input from users, extension developers, whatever. But "I use 10.4 and it would suck for me" is not really input; we _know_ there are people in that position.
> you should at least listen to the wishes of the users
Heck, I'm not just listening to what you're saying, I'm even responding to it. There's a difference between "listening to X" and "doing whatever X wants".
> IMO a good start toward fixing this would be to change the Bugzilla > etiquette guidelines so that under no circumstances short of actual > spam is it acceptable to tell anyone to stop commenting on a bug.
That's not necessarily workable. If X files a bug and Y tries along and tries to hijack it, I think it's perfectly reasonable to ask Y to file a separate bug (or file one) and ask Y to take the comments there.
>> On 2/7/2010 9:20 AM, Daniel Veditz wrote: >>> On 2/6/10 11:16 AM, Phillip Jones wrote: >>>> In in the Past I tested nightly until I figured what I said about >>>> bugs or suggestion were laughed at or ignored. since users don't >>>> know anything.
>>> I'm sorry you had that experience -- no one should be laughing at >>> anyone trying to give good feedback.
>> But it is common nevertheless. And 'ignored' is even more common.
> Yup. And it only takes one bad bugzilla (or support forum) experience > to put a user off contacting us *forever*, & they will tell all their > friends not to bother trying to give us feedback, too.
> IMO a good start toward fixing this would be to change the Bugzilla > etiquette guidelines so that under no circumstances short of actual > spam is it acceptable to tell anyone to stop commenting on a bug.
There's a middle ground between "reject all feedback" and "will accept endless advocacy comments." I agree we don't always handle negative responses well, but newsgroups and other avenues other than our task management system exist to handle feedback and discussion. Continued venting/anger in bugs does not help anyone, and allowing it will not help us in any measurable or imaginary way.
Boris Zbarsky <bzbar...@mit.edu> wrote: > On 2/7/10 2:29 PM, Zack Weinberg wrote: > > IMO a good start toward fixing this would be to change the Bugzilla > > etiquette guidelines so that under no circumstances short of actual > > spam is it acceptable to tell anyone to stop commenting on a bug.
> That's not necessarily workable. If X files a bug and Y tries along > and tries to hijack it, I think it's perfectly reasonable to ask Y to > file a separate bug (or file one) and ask Y to take the comments > there.
There's a huge difference between "This bug is about A, not B; please (file a new bug for B | discuss B in bug NNNNN, which is about that)" and "Stop complaining about this bug, it'll get fixed when it gets fixed". The latter is what I think we should stop doing, *even though* it means more nagging from bugmail.
Mike Connor <mcon...@mozilla.com> wrote: > On 7-Feb-10, at 8:29 PM, Zack Weinberg wrote: > > IMO a good start toward fixing this would be to change the Bugzilla > > etiquette guidelines so that under no circumstances short of actual > > spam is it acceptable to tell anyone to stop commenting on a bug.
> There's a middle ground between "reject all feedback" and "will > accept endless advocacy comments."
From our perspective, yes.
From the perspective of each user who has gone to the trouble of creating a bugzilla account so they can tell us that (insert your favorite old bug here) is a problem for them too, only to be told to go away, there is no difference.
Because of that, I think we need to permit, and possibly even respond positively to, advocacy and venting in bugs EVEN THOUGH it is unhelpful to the small group of people who use bugzilla as a task queue.
Being ignored is arguably even worse than being told to go away. We should be taking efforts up to and including hiring more support staff in order to ensure that no one is ignored. Support scales more easily than devs.
