Announcing end of support for operating systems (was: Re: ARMv6)
Henri Sivonen
> Fennec still supports Android 2.0 (check out AndroidManifest.xml.in), we
> just stopped talking about Android 2.0 cause it's so darn old and
> insignificant.
>
> Dropping Maemo 4 and 5 were discussed during their respective cycles. Since
> neither was ever a tier 1 platform, we didn't feel the need to
> over-communicate.
With my user hat on, I always feel Mozilla *under*-communicates EOLs
whether it be the security EOL of a Mozilla branch or the end of
support for Mozilla products on a given OS version.
I read Mozilla communications more than users who don't also work on
Gecko, but as a user of Fennec on Maemo 4, I had no warning of product
EOL before a build that no longer worked was pushed to me through
updates. (To be clear, from the point of view of Mozilla's resource
allocation, I think it made good sense to drop support for Maemo 4.)
As would-be user of Android, I could notice that the system
requirements had changed from 2.0 to 2.1 minimum, but I saw no
announcement. Maybe from the point of view of someone who makes the
call whether to post on an official blog about the EOL of a system
that few people use anymore, it seems like making a too big a deal
about something that is somewhat of a negative thing to announce and
touches few people and, therefore, not worth drawing attention to. But
as a user, at least I get the feeling that my systems might be left
unsupported unpredictably when the track record is that end of support
for various systems isn't announced on user-facing communication
channels ahead of time.
This reminds me: does the "Firefox is up to date" indicator have the
capability to say "Firefox is not up to date but an up-to-date version
is not available for your system"?
--
Henri Sivonen
hsiv...@iki.fi
http://hsivonen.iki.fi/
Matt Brubeck
> With my user hat on, I always feel Mozilla *under*-communicates EOLs
> whether it be the security EOL of a Mozilla branch or the end of
> support for Mozilla products on a given OS version.
This is going to become more important with rapid releases. When
Firefox 4 dropped support for PowerPC/10.4 Macs (which was definitely
under-communicated and caused major hassles for users who deleted
Firefox 3.6 only to find that Firefox 4 wouldn't run), at least those
users still received security updates for Firefox 3.6.
But if Firefox 10 were to drop support for any hardware or OS platform,
users would instantly be stranded on an insecure version, and we should
warn them to migrate to a different browser immediately (if there is any
supported browser available for that platform).
> I read Mozilla communications more than users who don't also work on
> Gecko, but as a user of Fennec on Maemo 4, I had no warning of product
> EOL before a build that no longer worked was pushed to me through
> updates. (To be clear, from the point of view of Mozilla's resource
> allocation, I think it made good sense to drop support for Maemo 4.)
The upcoming demotion of Maemo 5 to a tier-3 platform will be our next
test. Let's see how much better we can do!
> As would-be user of Android, I could notice that the system
> requirements had changed from 2.0 to 2.1 minimum, but I saw no
> announcement.
To clarify this one point again: We have not changed (and will not
change) our system requirements from Android 2.0 to Android 2.1. For
some reason someone changed the text on some web pages, but Firefox is
still compatible with Android 2.0. When the minimum version changes in
the future, it will be from 2.0 to 2.2.
This is somewhat academic since the only a handful of devices ever
shipped with Android 2.0; all except some specialized e-book readers
were updated to 2.1 a few months after release; and for other reasons
none of them are supported by Firefox. Google doesn't even report 2.0
usage any more; when they last reported it about 12 months ago, it was
already down to 0.1% of Android devices:
http://developer.android.com/resources/dashboard/platform-versions.html
(Use archive.org to see older versions of the report.)
Matt Brubeck
> whether it be the security EOL of a Mozilla branch or the end of
> support for Mozilla products on a given OS version.
This is going to become more important with rapid releases. When
Firefox 4 dropped support for PowerPC/10.4 Macs (which was definitely
under-communicated and caused major hassles for users who deleted
Firefox 3.6 only to find that Firefox 4 wouldn't run), at least those
users still received security updates for Firefox 3.6.
But if Firefox 10 were to drop support for any hardware or OS platform,
users would instantly be stranded on an insecure version, and we should
warn them to migrate to a different browser immediately (if there is any
supported browser available for that platform).
> I read Mozilla communications more than users who don't also work on
> Gecko, but as a user of Fennec on Maemo 4, I had no warning of product
> EOL before a build that no longer worked was pushed to me through
> updates. (To be clear, from the point of view of Mozilla's resource
> allocation, I think it made good sense to drop support for Maemo 4.)
The upcoming demotion of Maemo 5 to a tier-3 platform will be our next
test. Let's see how much better we can do!
> As would-be user of Android, I could notice that the system
> requirements had changed from 2.0 to 2.1 minimum, but I saw no
> announcement.
To clarify this one point again: We have not changed (and will not
Michael Lefevre
> On 09/01/2011 11:31 AM, Henri Sivonen wrote:
>> With my user hat on, I always feel Mozilla *under*-communicates EOLs
>> whether it be the security EOL of a Mozilla branch or the end of
>> support for Mozilla products on a given OS version.
>
> This is going to become more important with rapid releases. When Firefox
> 4 dropped support for PowerPC/10.4 Macs (which was definitely
> under-communicated and caused major hassles for users who deleted
> Firefox 3.6 only to find that Firefox 4 wouldn't run), at least those
> users still received security updates for Firefox 3.6.
> The upcoming demotion of Maemo 5 to a tier-3 platform will be our next
> test. Let's see how much better we can do!
I'm not sure it's entirely relevant, but I thought it might be worth
mentioning https://bugzilla.mozilla.org/show_bug.cgi?id=666735 (Don't
offer a Firefox 7 update to users with a too-old version of Freetype),
which is going to leave users on older Linux distros on Firefox 6
(although I guess they can make it work if they upgrade some libraries -
that's not entirely clear to me)
Michael
timeless
the browser is unsupported and pointing to a kb page where people can
offer suggested alternatives.
Personally, I'd love to see a cloud renderer as an alternative for old
platforms. I was hoping SkyFire would be that.
On 9/1/11, Henri Sivonen <hsiv...@iki.fi> wrote:
> On Thu, Sep 1, 2011 at 6:14 PM, Mark Finkle <mark....@gmail.com> wrote:
>> Fennec still supports Android 2.0 (check out AndroidManifest.xml.in), we
>> just stopped talking about Android 2.0 cause it's so darn old and
>> insignificant.
>>
>> Dropping Maemo 4 and 5 were discussed during their respective cycles.
>> Since
>> neither was ever a tier 1 platform, we didn't feel the need to
>> over-communicate.
>
> With my user hat on, I always feel Mozilla *under*-communicates EOLs
> whether it be the security EOL of a Mozilla branch or the end of
> support for Mozilla products on a given OS version.
>
> I read Mozilla communications more than users who don't also work on
> Gecko, but as a user of Fennec on Maemo 4, I had no warning of product
> EOL before a build that no longer worked was pushed to me through
> updates. (To be clear, from the point of view of Mozilla's resource
> allocation, I think it made good sense to drop support for Maemo 4.)
> As would-be user of Android, I could notice that the system
> requirements had changed from 2.0 to 2.1 minimum, but I saw no
> announcement. Maybe from the point of view of someone who makes the
> call whether to post on an official blog about the EOL of a system
> that few people use anymore, it seems like making a too big a deal
> about something that is somewhat of a negative thing to announce and
> touches few people and, therefore, not worth drawing attention to. But
> as a user, at least I get the feeling that my systems might be left
> unsupported unpredictably when the track record is that end of support
> for various systems isn't announced on user-facing communication
> channels ahead of time.
>
> This reminds me: does the "Firefox is up to date" indicator have the
> capability to say "Firefox is not up to date but an up-to-date version
> is not available for your system"?
>
> --
> Henri Sivonen
> hsiv...@iki.fi
> http://hsivonen.iki.fi/
> _______________________________________________
> dev-planning mailing list
> dev-pl...@lists.mozilla.org
> https://lists.mozilla.org/listinfo/dev-planning
>
--
Sent from my mobile device
Cameron Kaiser
> the browser is unsupported and pointing to a kb page where people can
> offer suggested alternatives.
>
> Personally, I'd love to see a cloud renderer as an alternative for old
> platforms. I was hoping SkyFire would be that.
I myself would like to see SkyFire resurrected. It would be really
nice to have access to such a thing for underpowered machines too. If
someone is resurrecting that officially, I would like to get involved.
In the "unsupported masses" category, I suspect that 10.5 will be the
next platform to get the axe, which is of great concern and interest
to us in TenFourFox-land as right now 10.5 compatibility gives us
"just close enough" compatibility to continue to hack in our necessary
10.4 support (for Classic and G3 users). When 10.5 support is dropped
and that code gradually erodes, if we're not able to compensate my
plan is to continue a "long term support" branch on whatever that last
version was and backport fixes (we almost had to do this back with Fx4
because of linker issues that one of our crack hackers was able to
resolve). However, I suspect that this will leave a number of regular
Fx users high and dry and perhaps that would be a logical point for an
official LTS branch if such a thing gained traction. I imagine 3.6's
lifetime not to be much longer than the end of the year, and it is so
far the de facto current LTS, so perhaps that's timely.
Cameron Kaiser
Henri Sivonen
> But if Firefox 10 were to drop support for any hardware or OS platform,
> users would instantly be stranded on an insecure version, and we should warn
> them to migrate to a different browser immediately (if there is any
> supported browser available for that platform).
Sync makes Firefox more sticky, which is nice for retaining market
share, but the flip side is that users who've gotten used to Sync are
likely be more unhappy about having to switch browsers to stay secure
than previously. (I sure was unhappy about having to switch to Opera
Mobile on my N800 after using Firefox Sync in Fennec on it--even
though Opera Mobile was more responsive, etc.)
The next desktop OSs to be dropped are most likely going to be Windows
2000 and Windows XP SP1. In the case of XP SP1 at least, I think it
would make more sense to let the user know that a free service pack
update is needed in order to update Firefox to stay safe.
The proportion of the userbase that is running Windows 2000 and XP SP1
still is probably very small. Yet, I predict that if those users are
abandoned without communication, it will breed even more negative
sentiment about rapid release *among people who aren't even running
Windows 2000 or XP SP1 but hear eventually about what happened to
someone else*. I think it would make sense to make sure that at the
latest the release train before the one that no longer runs on Windows
2000 and XP SP1 has the capability to avoid downloading and installing
the next train that won't run and to display (in advance!) relevant
advice to the users (instruct XP SP1 users to install a newer service
pack and instruct Windows 2000 users to get a new OS ASAP). (Easy for
me to say, since I don't work on the relevant code.)
(I suppose in principle it would be technically correct to instruct
Windows 2000 users to switch to Opera. However, simple saying "another
browser" to Windows 2000 users wouldn't help our remaining Windows
2000 users stay safe, because the better-known alternatives don't
apply: IE is no longer being patched for Windows 2000 and Chrome and
Safari aren't available.)
Does Firefox already have such communication preparations for the EOL
for Windows 2000 and XP SP1 support? Should I file a bug?
>> As would-be user of Android, I could notice that the system
>> requirements had changed from 2.0 to 2.1 minimum, but I saw no
>> announcement.
>
> To clarify this one point again: We have not changed (and will not change)
> our system requirements from Android 2.0 to Android 2.1. For some reason
> someone changed the text on some web pages, but Firefox is still compatible
> with Android 2.0. When the minimum version changes in the future, it will
> be from 2.0 to 2.2.
>
> This is somewhat academic since the only a handful of devices ever shipped
> with Android 2.0;
It might be academic as far as the actual user numbers go for Android
2.0. It's not academic as far as the perception of Mozilla's
policies/attitude goes. (Well, it could be academic if I am the only
person paying attention to stuff like this. I don't know how
representative I am.) When I noticed that the stated requirement
changed without announcement, I thought to myself that Firefox for
Android hasn't yet been established as a seriously supported product
and is still treated as something that can have its requirements
changed without warning. Now, that's the wrong conclusion when the
requirement haven't really changed, but before this thread that's what
I thought when I saw that the stated requirements changed silently so
soon after the initial release (4.0). Combined with the sad state of
Android OS updates, this affected my planning so that I chose to
postpone becoming a Firefox for Android user.
Kyle Huey
> The next desktop OSs to be dropped are most likely going to be Windows
> 2000 and Windows XP SP1.
Yes, this is likely the case.
> In the case of XP SP1 at least, I think it
> would make more sense to let the user know that a free service pack
> update is needed in order to update Firefox to stay safe.
>
Yes, telling users this might be a good idea. My understanding is that a
lot of the people on ancient XP service packs are there because they have
pirated copies and the newer service packs have anti-piracy features in
them, so I don't know how effective this will be.
> I think it would make sense to make sure that at the
> latest the release train before the one that no longer runs on Windows
> 2000 and XP SP1 has the capability to avoid downloading and installing
> the next train that won't run
This is one of two things blocking dropping support for these operating
systems.
> and to display (in advance!) relevant
> advice to the users (instruct XP SP1 users to install a newer service
> pack and instruct Windows 2000 users to get a new OS ASAP). (Easy for
> me to say, since I don't work on the relevant code.)
>
That we haven't planned to do. The current set of bugs on file would result
in us just not offering these users updates past the last release that
supports their machine.
(I suppose in principle it would be technically correct to instruct
> Windows 2000 users to switch to Opera. However, simple saying "another
> browser" to Windows 2000 users wouldn't help our remaining Windows
> 2000 users stay safe, because the better-known alternatives don't
> apply: IE is no longer being patched for Windows 2000 and Chrome and
> Safari aren't available.)
>
If people are still on Win2k (unless they have some sort of support contract
with Microsoft) their systems are almost certainly horribly insecure
(security fixes stopped being issued some time ago), so I think that "you
need to move to a newer operating system" is much better advice than "you
need to switch to browser X, as they're the only ones who still support your
system". This might just be my gut reaction to the idea of telling our
users to switch to a different browser though.
For those who do have support contracts with Microsoft, I would expect (but
I certainly don't know) that that contract includes continued security fixes
for IE 6.
Does Firefox already have such communication preparations for the EOL
> for Windows 2000 and XP SP1 support? Should I file a bug?
>
Filing some bugs would be useful here. The "there's an update to Firefox
but it won't run because your system is too old" idea is definitely worth a
bug.
- Kyle
Henri Sivonen
> Filing some bugs would be useful here. The "there's an update to Firefox
> but it won't run because your system is too old" idea is definitely worth a
> bug.
OK. Filed bug https://bugzilla.mozilla.org/show_bug.cgi?id=684227
Robert Kaiser
> This reminds me: does the "Firefox is up to date" indicator have the
> capability to say "Firefox is not up to date but an up-to-date version
> is not available for your system"?
No, we always show "up to date" if nothing is found, even if we can't
access the update server or something. We tell people they are secure
and current if they are unsupported or the connection to the update
server is impaired, e.g. through a MITM blocking your access to the
update server. I guess we just fear telling the user we needs to do
something himself which we can't do automatically at that time. I filed
a bug, but was told that's intentional design.
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! :)
Henri Sivonen
> No, we always show "up to date" if nothing is found, even if we can't access
> the update server or something. We tell people they are secure and current
> if they are unsupported or the connection to the update server is impaired,
> e.g. through a MITM blocking your access to the update server. I guess we
> just fear telling the user we needs to do something himself which we can't
> do automatically at that time. I filed a bug, but was told that's
> intentional design.
Wow. That's really bad (both what happens and the position that it's
by design). :-(
I suspected this might be the case (after noticing that my
self-compiled Firefox claimed to be up-to-date) and was about to file
a bug. My Bugzilla skills seem to be weak. What's the bug # for the
bug you filed?
Jesper Kristensen
> Henri Sivonen schrieb:
>> This reminds me: does the "Firefox is up to date" indicator have the
>> capability to say "Firefox is not up to date but an up-to-date version
>> is not available for your system"?
>
> No, we always show "up to date" if nothing is found, even if we can't
> access the update server or something. We tell people they are secure
> and current if they are unsupported or the connection to the update
> server is impaired, e.g. through a MITM blocking your access to the
> update server. I guess we just fear telling the user we needs to do
> something himself which we can't do automatically at that time. I filed
> a bug, but was told that's intentional design.
>
> Robert Kaiser
If that is true, that is kind of ironic. Within the discussion about the
removal of the version number from the about box, one statement was that
many users didn't trust that automatic update systems would tell the
truth. Someone from the UX team said it was an important goal to design
the update system such that the user would trust it.
Daniel Veditz
> No, we always show "up to date" if nothing is found, even if we
> can't access the update server or something. We tell people they are
> secure and current if they are unsupported or the connection to the
> update server is impaired, e.g. through a MITM blocking your access
> to the update server.
Part of the bug for getting removing version numbers from the about
box talked about updating the text to saying something like "You're
up to date, last successful check 7 hours ago" or something along
those lines. I hope that part makes it into the product. Although if
it says you're up to date simply because we dropped support that's
wrong.
Axel Hecht
I quite honestly hope that it doesn't. Whenever you open that dialog,
you check for updates. Thus, showing when you checked is rather pointless.
For people close to the update server, the update check progress might
be too quick for them to notice what happens, that's something worth fixing.
The other aspect of that ui is the different scenarios in which it's not
just "OK" or "update available" like disabled, no permissions, no
updates for that platform.
Axel
Daniel Veditz
> I quite honestly hope that it doesn't. Whenever you open that
> dialog, you check for updates. Thus, showing when you checked is
> rather pointless.
If things are working OK. The explanation in the bug was that the
automated updates checks can fail for various reasons and that would
be a place to tell users about that. Without a "Check now" button
nearby I would think that's just going to confuse and possibly
infuriate users.
-Dan Veditz