5.0.1 for Windows/Linux will be also released ?
There's no plan to do that at the moment. 5.0.1 for Mac fixes a
specific problem where we're hitting a bug in the system font APIs on OS
X 10.7. Otherwise it's identical to 5.0. The relevant code is not used
at all on Windows/Linux.
-Boris
No idea. If I had to bet, I'd bet because it was all done with a script
and the script Just Does That. But that's a total guess.
-Boris
And this guess is correct. It's just easier to tell the system to kick
out a full round of builds than to manually do one platform's builds.
Those other builds will not be shipped.
Chris Blizzard has a post coming about this probably tomorrow.
- A
> _______________________________________________
> dev-planning mailing list
> dev-pl...@lists.mozilla.org
> https://lists.mozilla.org/listinfo/dev-planning
No. They are not released. Putting builds on FTP is not a release. The
Windows and Linux builds have absolutely no changes in them and will not
be released. They exist because it's easier to let the automation make
them than to build Mac by hand.
- A
I am wrong. They are released. They are not offered as updates to
Firefox 5 users except on Mac though.
New users coming to the site to download Firefox for the first time will
get 5.0.1 which is identical to 5.0 on Windows and Linux in all but
version number.
- A
both are on official download page.
http://www.mozilla.com/en-US/firefox/all.html
http://www.mozilla.com/en-US/firefox/all-older.html
Why was 5.0.1 released at all for Windows and Linux? The current
situation seems to have at least the following possible downsides:
1) Windows and Linux users wasting time worrying about autoupdate not
working if they happen to see that 5.0.1 is offered for new installs.
2) Windows and Linux users wasting time manually downloading and
installing 5.0.1 if they notice it is available for download.
3) Since the point release shows in the UA string, Windows and Linux
users being slightly more fingerprintable as sites are able to
distinguish users who've manually downloaded Firefox after a particular
day.
What's the upside? Is pushing the same point release for all platforms
to the Web site also part of the automation?
--
Henri Sivonen
hsiv...@iki.fi
http://hsivonen.iki.fi/
> On Mon, 2011-07-11 at 23:03 -0700, Asa Dotzler wrote:
>> New users coming to the site to download Firefox for the first time
>> will
>> get 5.0.1 which is identical to 5.0 on Windows and Linux in all but
>> version number.
>
> Why was 5.0.1 released at all for Windows and Linux? The current
> situation seems to have at least the following possible downsides:
>
> 1) Windows and Linux users wasting time worrying about autoupdate not
> working if they happen to see that 5.0.1 is offered for new installs.
Who cares, they will manually install 5.0.1 and be fine, or be worried and still be secure. Non-issue, and it will all be over on August 16th.
> 2) Windows and Linux users wasting time manually downloading and
> installing 5.0.1 if they notice it is available for download.
No one will notice this unless they are explicitly looking for it. I bet it's less than .001% that would waste time, versus a greater percentage that would waste time downloading in the background and installing if we offered it to all platforms. Non-issue, and it will all be over on August 16th.
> 3) Since the point release shows in the UA string, Windows and Linux
> users being slightly more fingerprintable as sites are able to
> distinguish users who've manually downloaded Firefox after a particular
> day.
Who cares? This is only a concern until August 16th and if those users want to fingerprint themselves by downloading an unadvertised build, so be it. They have to explicitly search out the manual version...
> What's the upside? Is pushing the same point release for all platforms
> to the Web site also part of the automation?
a) We didn't want Linux and Windows users to download the 5.0.1 update when they had no benefit. Why waste their bandwidth?
b) We have no way of saying "5.0.1 is latest for Mac but 5.0 is latest for other platforms" on the website. We could have hacked it together but were worried about downstream projects (like input.mozilla.org, crash-stats, etc). It seemed silly to not update the website with the latest version, as QA had qualified the updates across all versions and we didn't want Mac users to download 5.0 and then immediately get an update to 5.0.1.
Though this release situation was not ideal we feel it was in the best interests of the majority of our users.
Thanks,
Christian
It's not nice to make people worry needlessly even if it a relatively
small proportion of the user base.
> > 2) Windows and Linux users wasting time manually downloading and
> > installing 5.0.1 if they notice it is available for download.
>
> No one will notice this unless they are explicitly looking for it. I bet it's less than .001% that would waste time, versus a greater percentage that would waste time downloading in the background and installing if we offered it to all platforms. Non-issue, and it will all be over on August 16th.
(I wasn't implying any suggestion of pushing the update to Windows and
Linux users.)
> b) We have no way of saying "5.0.1 is latest for Mac but 5.0 is latest for other platforms" on the website.
I see. That explains the situation even though it's unfortunate that
that capability is lacking.
I suspect the negative "WTF? *Another* update?" response would have
been far more common and stronger had 5.0.1 been pushed for
Windows/Linux.
N
Or will decide that security updates are broken because they're not
getting it and switch browsers. I know at least one person who claims
to have done that when we announced that 4.0 was out of support, and a
day later they had still not gotten the update offer (due to the update
throttling on 5.0 we did).
A minority case for sure; I'm not sure whether it's worth investing in
what it would take to offer different versions to people on the major
platforms in the future if this situation comes up again to address this
case. But worth keeping in mind.
-Boris
I do, as 5.0 and 5.0.1 are showing up as two completely different things
in crash stats, and we can only offer partial updates from one of them
to 6, both of which are quite suboptimal...
> a) We didn't want Linux and Windows users to download the 5.0.1 update when they had no benefit.
Then we shouldn't offer them anywhere at all.
Robert Kaiser
--
Note that any statements of mine - no matter how passionate - are never
meant to be offensive but very often as food for thought or possible
arguments that we as a community should think about. And most of the
time, I even appreciate irony and fun! :)
On Jul 12, 2011, at 7:19 AM, Robert Kaiser <ka...@kairo.at> wrote:
> Christian Legnitto schrieb:
>> Who cares,
>
> I do, as 5.0 and 5.0.1 are showing up as two completely different things in crash stats, and we can only offer partial updates from one of them to 6, both of which are quite suboptimal...
Right, I explicitly asked the Socorro team if this was going to be a problem from the system side and was told it would not be in general.
We also made a conscious decision on the different "partial vs download" scenarios and are aware of the partial issue. I looped Asa in and we agreed on the various tradeoffs.
Sorry for making extra crash-stats work :-/
>
>> a) We didn't want Linux and Windows users to download the 5.0.1 update when they had no benefit.
>
> Then we shouldn't offer them anywhere at all.
Again, we chose to offer all for manual download because of the way product-details works. There were tradeoffs here as well.
>
There's no problem with Socorro there at all, that's right. What I meant
is that it now looks as if we need to track 5.0.1 for Windows (and
Linux) as well in terms of crash stats, which I didn't think we'd have
to, as this should be a Mac-only chemspill.
> Again, we chose to offer all for manual download because of the way product-details works. There were tradeoffs here as well.
And that decision is what causes me some grief - but I guess it's some
website automation issue. Well, we know now we need to monitor one
version more for all OSes and not just Mac, so that's at least good to know.
When a user chooses to click to update, they should get one, or else
they wonder why it exists on the download page and nothing happens
when they click. Then they call and question system builders about
things like this and in turn system builders replace products with
alternatives rather than have to deal with numerous phone calls.
Therefore this is by definition a marketing bug, but a bug
nonetheless. v5.0.1 doesn't have to be mass delivered, but why not
have the 'useless' changes listed on the official page and have the
'useless' number change switched when users choose to click on 'Check
for Updates'?
a) That's not the official page of what changed...that's where we post what security fixes were fixed. No security fixes were fixed in this version so it doesn't show there. For a list of changes, you look at release notes (https://www.mozilla.com/en-US/firefox/5.0.1/releasenotes/) which has:
*Worked around an issue in Mac OS X 10.7 that could cause Firefox to crash
*Worked around an issue caused by Apple's "Java for Mac OS X 10.6 Update 5" where the Java plugin would not be loaded
b) The announcement was a bit delayed (there was a rocky handoff entirely caused by me) but it was put up this morning on the Developer News blog @ https://developer.mozilla.org/devnews/index.php/2011/07/12/firefox-5-0-1-and-3-6-19-compatibility-updates-now-available/
c) I blogged a bit more about the reasoning behind shipping the updates only for Macs while allowing all platforms to download 5.0.1 from mozilla.com:
Hope that clarifies a bit.
Thanks,
Christian
> a) We didn't want Linux and Windows users to download the 5.0.1 update when they had no benefit. Why waste their bandwidth?
I think that, in the past, I would have argued here about consistent
versioning (across all platforms) being a win from the point of view of
(1) no user confusion when someone notices they have 5.0.0 and their OS
X friend having 5.0.1 and (2) taking less time to just Do It than to
argue about it here.
But, as the version number becomes less-and-less important and more
difficult for end users to notice, this issue will resolve itself
without us needing to do anything.
Justin
Mike
The reason is on the RelEng side, as release automation only creates
partials from one other version to the current one - everything else
needs manual steps.
> *Worked around an issue in Mac OS X 10.7 that could cause Firefox to crash
> *Worked around an issue caused by Apple's "Java for Mac OS X 10.6 Update 5" where the Java plugin would not be loaded
Speaking of Apple problems, is there a bug open for this?
http://support.mozilla.com/en-US/questions/836622
Their web site still doesn't work with Gecko 6.
Awesomebar to the rescue:
<https://bugzilla.mozilla.org/show_bug.cgi?id=657469> :)