Are we pushing Firefox 3.5 too hard?

[iwod]

It seems to me the current builds are not near shipping quality. And given the amount of P1 bugs still exist. May be Mozilla decide to rethink their schedule?

[pal-moz]

http://groups.google.com/group/mozilla. ... ae837076c#

[MichaelKohler]

why "not near shipping quality"? because of these few bugs?

[SlimShady]

It looks to me they work hard and still cant fix all bugs. Maybe they need more developers to speed it up.

[michaell522]

It seems to me the current builds are not near shipping quality. And given the amount of P1 bugs still exist. May be Mozilla decide to rethink their schedule?

They did rethink the schedule, and are still doing that. One can discuss whether it was the right thing to do (although there is not much point in looking back now), but they decided a several months ago to push more things into Firefox 3.5. Since then they have added the extra beta to the schedule (so these builds aren't supposed to be shipping quality, they are supposed to be still be beta quality, and after the beta is out, then they will move onto release candidates). Then several weeks ago they decided to delay this beta (and have now delayed it a little further than expected then) rather than doing a release and adding yet another beta release.

It looks to me they work hard and still cant fix all bugs. Maybe they need more developers to speed it up.

I don't think more developers in general would help. Having more developers would tend to mean working on more features and changes at the same time, so there would be more blocking bugs to fix as they work towards the release.

The issue in general is not really the speed of development, but the gaps between releases. If you have more frequent releases, there is more time spent on polishing up the releases and probably less gets done over the course of several releases. But if you have more time between releases, you can be more flexible and get more into each release, but people (those using final releases anyway) have to wait longer before they get to use the new stuff. They decided last year to go for more frequent releases, but as work has continued there seems to be a bit of a slip towards getting more done before releasing (but no more than before).

[the_dees]

I personally think they branched one milestone too early. I really appreciate all the fixes that already went into 1.9.2, but not having to land each patch twice could've gained some time. Of course, i don't know enough to really predict what would've happend.

Anyway, will there be a 5th beta or will the next milestone become RC1?

[michaell522]

I personally think they branched one milestone too early. I really appreciate all the fixes that already went into 1.9.2, but not having to land each patch twice could've gained some time. Of course, i don't know enough to really predict what would've happend.

That's probably true. But if I remember correctly, they added the extra milestone after branching...

Anyway, will there be a 5th beta or will the next milestone become RC1?

A few weeks ago there was a discussion about whether to delay beta 4 by 2-3 weeks or to go ahead and release it and add another quick beta 6 weeks later, and they decided to delay beta 4. So as of then, the plan would be to have RC1 next. But depending how things turn out, that could change, of course - Mozilla isn't the type of organisation to do anything drastic like dropping features at the last second or knowingly releasing something of poor quality with the intention of fixing it up in a .1 release (as some companies do for commercial/marketing reasons) if they can handle things just by delaying things a bit.

[MichaelKohler]

IMO delaying beta releases is good, because so a lot of user can test the new features. This makes Firefox a better product.

[the_dees]

That's probably true. But if I remember correctly, they added the extra milestone after branching...

On the other hand, Tracemonkey is also responisble for a good chunk of delay. The upvar landing and other relates fixed account for another part.

But depending how things turn out, that could change, of course - Mozilla isn't the type of organisation to do anything drastic like dropping features at the last second or knowingly releasing something of poor quality with the intention of fixing it up in a .1 release

Hm, I guess the result of broader Beta 4 testing will tell whether a 5th beta is necessary.

[michaell522]

That's probably true. But if I remember correctly, they added the extra milestone after branching...

On the other hand, Tracemonkey is also responisble for a good chunk of delay. The upvar landing and other relates fixed account for another part.

Well, my impression is that's mostly the same hand, rather than the other one The upvar landing was mostly responsible for the tracemonkey delay, and the tracemonkey delay was mostly responsible for the extra milestone (I say "mostly" - there were some other bits as well...).

Hm, I guess the result of broader Beta 4 testing will tell whether a 5th beta is necessary.

Indeed. The beta testing will indicate how much work is needed, and the amount of work that's needed will dictate how much more testing is needed, and therefore whether there should be another beta or just release candidates...

[_Dexter_]

the multi-platform nature of FF must also hinders things.

[geeknik]

I dunno, with the exception of the big Vista crash bug that popped up this week, the Branch builds have been pretty good. Lots of tracemonkey landings, awesomebar was sped up, lots of little fixes too. I think 3.5b4 is gonna be a good release. But I am concerned about some javascript performance issues that need to be fixed before 3.5 is actually released.

[Erunno]

There has been discussion if TraceMonkey should be dropped for 3.1/3.5 and instead be introduced with Firefox.next some time ago on the planning mailing list. This would have allowed 3.1 to be released a couple of months earlier and Tracemonkey with the subsequent release (i.e. 3.2). As one developer correctly stated: What's the difference if one releases TraceMonkey in October with 3.1 due to months of delays or release it in October with 3.2? This would have allowed other completed features to be pushed to the public early and Mozilla developers would receive more feedback while at the same time more complicated features like TraceMonkey and the <video> support would have more time to cook.

Hopefully that long delay is only a slip and Firefox will indeed use a short, iterative development cycles for future versions.

[Kernel Sanders]

Personally I think they are pushing themselves too hard. 3.0 isn't really an old product, yet they are pushing for 3.5 already.

I would be happy with 1 firefox release per year rather than two. It would give the devs more time and it would be less pressure all around.

[teoli2003]

Kernel, Firefox 3.5 will be out exactly one year after Firefox 3 (June): we are already mid-April and we haven't reached the RC phase, which will last at least one month.