The future of Linux32 builds and tests

classic Classic list List threaded Threaded
27 messages Options
12
Reply | Threaded
Open this post in threaded view
|

The future of Linux32 builds and tests

Kim Moir-2
We recently disabled Linux32 talos testing on all trees.

Going forward, the question remains, do we need to run Linux32 tests at all?  This would reduce the complexity of our configs, as well as significantly reduce our AWS bill. Do the linux32 test results provide significant differences from the ones we run on on linux64? I see three options going forward

1) Run linux32 builds and tests periodically
2) Turn off tests but run the builds and make it a Tier 2 platform
3) Turn off Linux32 builds + tests entirely

Your data-driven input is sought,

cheers,
Kim

Bugs of note
Disable all 32-bit linux testing
https://bugzilla.mozilla.org/show_bug.cgi?id=1209932
Do we need linux32 talos?
https://bugzilla.mozilla.org/show_bug.cgi?id=1204920
Disable linux32 talos testing and reimage machines for windows testing
https://bugzilla.mozilla.org/show_bug.cgi?id=1208449
_______________________________________________
dev-planning mailing list
[hidden email]
https://lists.mozilla.org/listinfo/dev-planning
Reply | Threaded
Open this post in threaded view
|

Re: The future of Linux32 builds and tests

Millwood
[hidden email] wrote:

> We recently disabled Linux32 talos testing on all trees.
>
> Going forward, the question remains, do we need to run Linux32 tests at all?  This would reduce the complexity of our configs, as well as significantly reduce our AWS bill. Do the linux32 test results provide significant differences from the ones we run on on linux64? I see three options going forward
>
> 1) Run linux32 builds and tests periodically
> 2) Turn off tests but run the builds and make it a Tier 2 platform
> 3) Turn off Linux32 builds + tests entirely
>
> Your data-driven input is sought,
>
> cheers,
> Kim
>
> Bugs of note
> Disable all 32-bit linux testing
> https://bugzilla.mozilla.org/show_bug.cgi?id=1209932
> Do we need linux32 talos?
> https://bugzilla.mozilla.org/show_bug.cgi?id=1204920
> Disable linux32 talos testing and reimage machines for windows testing
> https://bugzilla.mozilla.org/show_bug.cgi?id=1208449
>
"Upgrading" a 32 bit linux system to 64 is not an easy prospect so
please keep building at least product versions for 32 bits.
_______________________________________________
dev-planning mailing list
[hidden email]
https://lists.mozilla.org/listinfo/dev-planning
Reply | Threaded
Open this post in threaded view
|

Re: The future of Linux32 builds and tests

Daniel Holbert-3
In reply to this post by Kim Moir-2
On 10/08/2015 10:01 AM, [hidden email] wrote:
> Going forward, the question remains, do we need to run Linux32 tests at all?  This would reduce the complexity of our configs, as well as significantly reduce our AWS bill. Do the linux32 test results provide significant differences from the ones we run on on linux64? I see three options going forward
>
> 1) Run linux32 builds and tests periodically
> 2) Turn off tests but run the builds and make it a Tier 2 platform
> 3) Turn off Linux32 builds + tests entirely
>
> Your data-driven input is sought,

When the possibility of making Linux32 "Tier 2" was brought up in 2012
[1], cpearce brought up a good point [2] about data on Linux users:

"...if we *were* to consider dropping a platform to tier 2, we should
make that decision with data to back it up, which for Linux should also
include data regarding the Firefox x86/x64 split in the major distros,
since most roll their own Firefox packages which we don't track."

We should try to get data from some 3rd party (not our download tallies)
about the Firefox-on-Linux 32-bit vs. 64-bit user stats, before we make
a call about demoting Linux32 to Tier 2. (Maybe we already have such data?)

~Daniel

[1]
https://groups.google.com/d/msg/mozilla.dev.planning/ahyq19APxb4/tvkekYhQbhcJ
[2]
https://groups.google.com/d/msg/mozilla.dev.planning/ahyq19APxb4/KO-P4xnwvIYJ
_______________________________________________
dev-planning mailing list
[hidden email]
https://lists.mozilla.org/listinfo/dev-planning
Reply | Threaded
Open this post in threaded view
|

Re: The future of Linux32 builds and tests

Gijs Kruitbosch ("Hannibal")
In reply to this post by Kim Moir-2
On 08/10/2015 18:01, [hidden email] wrote:

> We recently disabled Linux32 talos testing on all trees.
>
> Going forward, the question remains, do we need to run Linux32 tests at all?  This would reduce the complexity of our configs, as well as significantly reduce our AWS bill. Do the linux32 test results provide significant differences from the ones we run on on linux64? I see three options going forward
>
> 1) Run linux32 builds and tests periodically
> 2) Turn off tests but run the builds and make it a Tier 2 platform
> 3) Turn off Linux32 builds + tests entirely
>
> Your data-driven input is sought,
>
> cheers,
> Kim
>
> Bugs of note
> Disable all 32-bit linux testing
> https://bugzilla.mozilla.org/show_bug.cgi?id=1209932
> Do we need linux32 talos?
> https://bugzilla.mozilla.org/show_bug.cgi?id=1204920
> Disable linux32 talos testing and reimage machines for windows testing
> https://bugzilla.mozilla.org/show_bug.cgi?id=1208449

Pending more data about how many people use 32-bit fx on linux distros,
we could conceivably do both (1) and turn off 32-bit *debug* testing
completely, as nobody really uses those as daily builds anyway, and most
any issue should get caught by the opt tests or the x64 debug ones?

~ Gijs
_______________________________________________
dev-planning mailing list
[hidden email]
https://lists.mozilla.org/listinfo/dev-planning
Reply | Threaded
Open this post in threaded view
|

Re: The future of Linux32 builds and tests

Daniel Holbert-3
In reply to this post by Daniel Holbert-3
On 10/08/2015 11:34 AM, Daniel Holbert wrote:
> On 10/08/2015 10:01 AM, [hidden email] wrote:
> We should try to get data from some 3rd party (not our download tallies)
> about the Firefox-on-Linux 32-bit vs. 64-bit user stats, before we make
> a call about demoting Linux32 to Tier 2. (Maybe we already have such data?)

Ah -- looking at the bugs that kmoir linked to, it looks like we do have
some data on this, from telemetry (which is opt-in in release builds, so
it's not perfect but it's something):

 https://bugzilla.mozilla.org/show_bug.cgi?id=1204920#c1

The numbers there show that ~40% of our Linux users are using the 32-bit
version. (on release & beta channels)  That's more than I was expecting.

Given that, I'd be uncomfortable with option (3) ("turn off Linux32
builds + tests entirely").  I'd also be uncomfortable with option (2)
(doing builds but no tests), because then 40% of our linux users would
be getting un-tested bits.  (Likely-working since they work on other
platforms, but who knows.)

Option (1) (running tests occasionally) seems doable though, and
depending on how infrequent "occasionally" is, it seems like it could
get us cost-savings that are close to those from option (2).

~Daniel
_______________________________________________
dev-planning mailing list
[hidden email]
https://lists.mozilla.org/listinfo/dev-planning
Reply | Threaded
Open this post in threaded view
|

Re: The future of Linux32 builds and tests

Mike Hommey
On Thu, Oct 08, 2015 at 12:42:11PM -0700, Daniel Holbert wrote:

> On 10/08/2015 11:34 AM, Daniel Holbert wrote:
> > On 10/08/2015 10:01 AM, [hidden email] wrote: We should try to
> > get data from some 3rd party (not our download tallies) about the
> > Firefox-on-Linux 32-bit vs. 64-bit user stats, before we make a call
> > about demoting Linux32 to Tier 2. (Maybe we already have such data?)
>
> Ah -- looking at the bugs that kmoir linked to, it looks like we do
> have some data on this, from telemetry (which is opt-in in release
> builds, so it's not perfect but it's something):
>
>  https://bugzilla.mozilla.org/show_bug.cgi?id=1204920#c1
>
> The numbers there show that ~40% of our Linux users are using the
> 32-bit version. (on release & beta channels)  That's more than I was
> expecting.
>
> Given that, I'd be uncomfortable with option (3) ("turn off Linux32
> builds + tests entirely").  I'd also be uncomfortable with option (2)
> (doing builds but no tests), because then 40% of our linux users would
> be getting un-tested bits.  (Likely-working since they work on other
> platforms, but who knows.)
>
> Option (1) (running tests occasionally) seems doable though, and
> depending on how infrequent "occasionally" is, it seems like it could
> get us cost-savings that are close to those from option (2).

Now the interesting question, is how many of those users are running on
64-bits systems, and are only using 32-bits builds because that's what
they downloaded back when mozilla.org didn't even offer the 64-bits
version, which brings to the additional question: would it be worth
forcing 32->64-bit upgrades of Firefox.

Mike
_______________________________________________
dev-planning mailing list
[hidden email]
https://lists.mozilla.org/listinfo/dev-planning
Reply | Threaded
Open this post in threaded view
|

Re: The future of Linux32 builds and tests

William Lachance-2
In reply to this post by Daniel Holbert-3
On 2015-10-08 5:26 PM, Mike Hommey wrote:

> On Thu, Oct 08, 2015 at 12:42:11PM -0700, Daniel Holbert wrote:
>> On 10/08/2015 11:34 AM, Daniel Holbert wrote:
>>   https://bugzilla.mozilla.org/show_bug.cgi?id=1204920#c1
>>
>> The numbers there show that ~40% of our Linux users are using the
>> 32-bit version. (on release & beta channels)  That's more than I was
>> expecting.
>>
>> Given that, I'd be uncomfortable with option (3) ("turn off Linux32
>> builds + tests entirely").  I'd also be uncomfortable with option (2)
>> (doing builds but no tests), because then 40% of our linux users would
>> be getting un-tested bits.  (Likely-working since they work on other
>> platforms, but who knows.)
>>
>> Option (1) (running tests occasionally) seems doable though, and
>> depending on how infrequent "occasionally" is, it seems like it could
>> get us cost-savings that are close to those from option (2).
>
> Now the interesting question, is how many of those users are running on
> 64-bits systems, and are only using 32-bits builds because that's what
> they downloaded back when mozilla.org didn't even offer the 64-bits
> version, which brings to the additional question: would it be worth
> forcing 32->64-bit upgrades of Firefox.

Aren't the vast majority of users running a version of Firefox
distributed by their vendor (of which Ubuntu probably has the largest
share, see e.g.
http://stats.wikimedia.org/wikimedia/squids/SquidReportOperatingSystems.htm)?
I don't think any metrics gathered from the Firefox builds which we
distribute are going to be very useful for measuring the impact of this
change.

I'd suggest reaching out to at least Ubuntu, Debian, and Fedora on this
and see what they think, at least as input. They're the domain experts,
and we really depend on their goodwill for ensuring that Firefox remains
the default browser for their operating system.

Will
_______________________________________________
dev-planning mailing list
[hidden email]
https://lists.mozilla.org/listinfo/dev-planning
Reply | Threaded
Open this post in threaded view
|

Re: The future of Linux32 builds and tests

Mike Hommey
On Thu, Oct 08, 2015 at 05:43:48PM -0400, William Lachance wrote:

> On 2015-10-08 5:26 PM, Mike Hommey wrote:
> >On Thu, Oct 08, 2015 at 12:42:11PM -0700, Daniel Holbert wrote:
> >>On 10/08/2015 11:34 AM, Daniel Holbert wrote:
> >>  https://bugzilla.mozilla.org/show_bug.cgi?id=1204920#c1
> >>
> >>The numbers there show that ~40% of our Linux users are using the
> >>32-bit version. (on release & beta channels)  That's more than I was
> >>expecting.
> >>
> >>Given that, I'd be uncomfortable with option (3) ("turn off Linux32
> >>builds + tests entirely").  I'd also be uncomfortable with option (2)
> >>(doing builds but no tests), because then 40% of our linux users would
> >>be getting un-tested bits.  (Likely-working since they work on other
> >>platforms, but who knows.)
> >>
> >>Option (1) (running tests occasionally) seems doable though, and
> >>depending on how infrequent "occasionally" is, it seems like it could
> >>get us cost-savings that are close to those from option (2).
> >
> >Now the interesting question, is how many of those users are running on
> >64-bits systems, and are only using 32-bits builds because that's what
> >they downloaded back when mozilla.org didn't even offer the 64-bits
> >version, which brings to the additional question: would it be worth
> >forcing 32->64-bit upgrades of Firefox.
>
> Aren't the vast majority of users running a version of Firefox distributed
> by their vendor (of which Ubuntu probably has the largest share, see e.g.
> http://stats.wikimedia.org/wikimedia/squids/SquidReportOperatingSystems.htm)?

That's the question I'd like to see answered the most. ISTR having been
said that a surprising large number of people are using our builds, but
I'm not sure I'm not making things up, and I don't know how large.

Mike
_______________________________________________
dev-planning mailing list
[hidden email]
https://lists.mozilla.org/listinfo/dev-planning
Reply | Threaded
Open this post in threaded view
|

Re: The future of Linux32 builds and tests

Daniel Holbert-3
In reply to this post by William Lachance-2
On 10/08/2015 02:43 PM, William Lachance wrote:
> Aren't the vast majority of users running a version of Firefox
> distributed by their vendor (of which Ubuntu probably has the largest
> share, see e.g.
> http://stats.wikimedia.org/wikimedia/squids/SquidReportOperatingSystems.htm)?
> I don't think any metrics gathered from the Firefox builds which we
> distribute are going to be very useful for measuring the impact of this
> change.

Oh, right -- it looks like telemetry isn't enabled in my Ubuntu-provided
Firefox build.  (The telemetry checkbox shown in the "how to enable
telemetry" article[1] just isn't present.)  I assumed it would be
enabled (and opt-in for users), but I assumed wrong.

So, it seems we aren't getting telemetry from Ubuntu users at least. So,
unless we're getting platform info some other way (AUS pings?), sounds
like we can't be sure about 32-vs-64 market share without talking to
distros.

~Dnaiel

[1]
https://support.mozilla.org/en-US/kb/share-telemetry-data-mozilla-help-improve-firefox
_______________________________________________
dev-planning mailing list
[hidden email]
https://lists.mozilla.org/listinfo/dev-planning
Reply | Threaded
Open this post in threaded view
|

Re: The future of Linux32 builds and tests

Lawrence Mandel-2
In reply to this post by William Lachance-2
On Thu, Oct 8, 2015 at 5:43 PM, William Lachance <[hidden email]>
wrote:

> On 2015-10-08 5:26 PM, Mike Hommey wrote:
>
>> On Thu, Oct 08, 2015 at 12:42:11PM -0700, Daniel Holbert wrote:
>>
>>> On 10/08/2015 11:34 AM, Daniel Holbert wrote:
>>>   https://bugzilla.mozilla.org/show_bug.cgi?id=1204920#c1
>>>
>>> The numbers there show that ~40% of our Linux users are using the
>>> 32-bit version. (on release & beta channels)  That's more than I was
>>> expecting.
>>>
>>> Given that, I'd be uncomfortable with option (3) ("turn off Linux32
>>> builds + tests entirely").  I'd also be uncomfortable with option (2)
>>> (doing builds but no tests), because then 40% of our linux users would
>>> be getting un-tested bits.  (Likely-working since they work on other
>>> platforms, but who knows.)
>>>
>>> Option (1) (running tests occasionally) seems doable though, and
>>> depending on how infrequent "occasionally" is, it seems like it could
>>> get us cost-savings that are close to those from option (2).
>>>
>>
>> Now the interesting question, is how many of those users are running on
>> 64-bits systems, and are only using 32-bits builds because that's what
>> they downloaded back when mozilla.org didn't even offer the 64-bits
>> version, which brings to the additional question: would it be worth
>> forcing 32->64-bit upgrades of Firefox.
>>
>
> Aren't the vast majority of users running a version of Firefox distributed
> by their vendor (of which Ubuntu probably has the largest share, see e.g.
> http://stats.wikimedia.org/wikimedia/squids/SquidReportOperatingSystems.htm)?
> I don't think any metrics gathered from the Firefox builds which we
> distribute are going to be very useful for measuring the impact of this
> change.
>
> I'd suggest reaching out to at least Ubuntu, Debian, and Fedora on this
> and see what they think, at least as input. They're the domain experts, and
> we really depend on their goodwill for ensuring that Firefox remains the
> default browser for their operating system.
>

Sylvestre - Can you get input from the Debian community about this?

Lawrence
_______________________________________________
dev-planning mailing list
[hidden email]
https://lists.mozilla.org/listinfo/dev-planning
Reply | Threaded
Open this post in threaded view
|

Re: The future of Linux32 builds and tests

danderson
In reply to this post by Kim Moir-2
On Thursday, October 8, 2015 at 10:02:02 AM UTC-7, [hidden email] wrote:

> We recently disabled Linux32 talos testing on all trees.
>
> Going forward, the question remains, do we need to run Linux32 tests at all?  This would reduce the complexity of our configs, as well as significantly reduce our AWS bill. Do the linux32 test results provide significant differences from the ones we run on on linux64? I see three options going forward
>
> 1) Run linux32 builds and tests periodically
> 2) Turn off tests but run the builds and make it a Tier 2 platform
> 3) Turn off Linux32 builds + tests entirely
>
> Your data-driven input is sought,
>
> cheers,
> Kim
>
> Bugs of note
> Disable all 32-bit linux testing
> https://bugzilla.mozilla.org/show_bug.cgi?id=1209932
> Do we need linux32 talos?
> https://bugzilla.mozilla.org/show_bug.cgi?id=1204920
> Disable linux32 talos testing and reimage machines for windows testing
> https://bugzilla.mozilla.org/show_bug.cgi?id=1208449

If AWS cost is a big concern, I wouldn't look at Linux 32 usage versus Linux 64 usage, but Linux versus other platforms. According to Telemetry, Linux has about 0.2% market share on Desktop, which means Windows has around 500X more users. Given that, it seems like we should not give Linux a large slice of infrastructure time, outside of the internal analysis builds like ASAN.
_______________________________________________
dev-planning mailing list
[hidden email]
https://lists.mozilla.org/listinfo/dev-planning
Reply | Threaded
Open this post in threaded view
|

Re: The future of Linux32 builds and tests

Mike Hommey
In reply to this post by Daniel Holbert-3
On Thu, Oct 08, 2015 at 06:01:50PM -0700, Daniel Holbert wrote:

> On 10/08/2015 02:43 PM, William Lachance wrote:
> > Aren't the vast majority of users running a version of Firefox
> > distributed by their vendor (of which Ubuntu probably has the largest
> > share, see e.g.
> > http://stats.wikimedia.org/wikimedia/squids/SquidReportOperatingSystems.htm)?
> > I don't think any metrics gathered from the Firefox builds which we
> > distribute are going to be very useful for measuring the impact of this
> > change.
>
> Oh, right -- it looks like telemetry isn't enabled in my Ubuntu-provided
> Firefox build.  (The telemetry checkbox shown in the "how to enable
> telemetry" article[1] just isn't present.)  I assumed it would be
> enabled (and opt-in for users), but I assumed wrong.

Telemetry is not enabled on our release builds either anyways.

Mike
_______________________________________________
dev-planning mailing list
[hidden email]
https://lists.mozilla.org/listinfo/dev-planning
Reply | Threaded
Open this post in threaded view
|

Re: The future of Linux32 builds and tests

Mike Hommey
In reply to this post by Lawrence Mandel-2
On Thu, Oct 08, 2015 at 09:08:04PM -0400, Lawrence Mandel wrote:

> On Thu, Oct 8, 2015 at 5:43 PM, William Lachance <[hidden email]>
> wrote:
>
> > On 2015-10-08 5:26 PM, Mike Hommey wrote:
> >
> >> On Thu, Oct 08, 2015 at 12:42:11PM -0700, Daniel Holbert wrote:
> >>
> >>> On 10/08/2015 11:34 AM, Daniel Holbert wrote:
> >>>   https://bugzilla.mozilla.org/show_bug.cgi?id=1204920#c1
> >>>
> >>> The numbers there show that ~40% of our Linux users are using the
> >>> 32-bit version. (on release & beta channels)  That's more than I was
> >>> expecting.
> >>>
> >>> Given that, I'd be uncomfortable with option (3) ("turn off Linux32
> >>> builds + tests entirely").  I'd also be uncomfortable with option (2)
> >>> (doing builds but no tests), because then 40% of our linux users would
> >>> be getting un-tested bits.  (Likely-working since they work on other
> >>> platforms, but who knows.)
> >>>
> >>> Option (1) (running tests occasionally) seems doable though, and
> >>> depending on how infrequent "occasionally" is, it seems like it could
> >>> get us cost-savings that are close to those from option (2).
> >>>
> >>
> >> Now the interesting question, is how many of those users are running on
> >> 64-bits systems, and are only using 32-bits builds because that's what
> >> they downloaded back when mozilla.org didn't even offer the 64-bits
> >> version, which brings to the additional question: would it be worth
> >> forcing 32->64-bit upgrades of Firefox.
> >>
> >
> > Aren't the vast majority of users running a version of Firefox distributed
> > by their vendor (of which Ubuntu probably has the largest share, see e.g.
> > http://stats.wikimedia.org/wikimedia/squids/SquidReportOperatingSystems.htm)?
> > I don't think any metrics gathered from the Firefox builds which we
> > distribute are going to be very useful for measuring the impact of this
> > change.
> >
> > I'd suggest reaching out to at least Ubuntu, Debian, and Fedora on this
> > and see what they think, at least as input. They're the domain experts, and
> > we really depend on their goodwill for ensuring that Firefox remains the
> > default browser for their operating system.
> >
>
> Sylvestre - Can you get input from the Debian community about this?

Debian is likely not very indicative of the overall Linux world. Taking
the primarily available figures from Debian and Ubuntu give completely
opposite pictures:
debian: i386: 51859, amd64: 132362
ubuntu: i386: 2065695, amd64: 673565

(from popcon.ubuntu.com and popcon.debian.org)

Mike
_______________________________________________
dev-planning mailing list
[hidden email]
https://lists.mozilla.org/listinfo/dev-planning
Reply | Threaded
Open this post in threaded view
|

Re: The future of Linux32 builds and tests

Mike Hommey
In reply to this post by danderson
On Thu, Oct 08, 2015 at 06:07:02PM -0700, [hidden email] wrote:

> On Thursday, October 8, 2015 at 10:02:02 AM UTC-7, [hidden email]
> wrote:
> > We recently disabled Linux32 talos testing on all trees.
> >
> > Going forward, the question remains, do we need to run Linux32 tests
> > at all?  This would reduce the complexity of our configs, as well as
> > significantly reduce our AWS bill. Do the linux32 test results
> > provide significant differences from the ones we run on on linux64?
> > I see three options going forward
> >
> > 1) Run linux32 builds and tests periodically 2) Turn off tests but
> > run the builds and make it a Tier 2 platform 3) Turn off Linux32
> > builds + tests entirely
> >
> > Your data-driven input is sought,
> >
> > cheers, Kim
> >
> > Bugs of note Disable all 32-bit linux testing
> > https://bugzilla.mozilla.org/show_bug.cgi?id=1209932 Do we need
> > linux32 talos?  https://bugzilla.mozilla.org/show_bug.cgi?id=1204920
> > Disable linux32 talos testing and reimage machines for windows
> > testing https://bugzilla.mozilla.org/show_bug.cgi?id=1208449
>
> If AWS cost is a big concern, I wouldn't look at Linux 32 usage versus
> Linux 64 usage, but Linux versus other platforms. According to
> Telemetry, Linux has about 0.2% market share on Desktop, which means
> Windows has around 500X more users. Given that, it seems like we
> should not give Linux a large slice of infrastructure time, outside of
> the internal analysis builds like ASAN.

I'd argue that if we are concerned by AWS cost, we should instead
(additionally?) consider consolidating B2G builds. There is no reason
we can't:
- stop building the whole android base (gonk) for every push.
- build one opt gecko and one debug one instead of one per gonk version.

Mike
_______________________________________________
dev-planning mailing list
[hidden email]
https://lists.mozilla.org/listinfo/dev-planning
Reply | Threaded
Open this post in threaded view
|

Re: The future of Linux32 builds and tests

mhoye
In reply to this post by danderson
On 2015-10-08 9:07 PM, [hidden email] wrote:
> If AWS cost is a big concern, I wouldn't look at Linux 32 usage versus
> Linux 64 usage, but Linux versus other platforms. According to
> Telemetry, Linux has about 0.2% market share on Desktop, which means
> Windows has around 500X more users
A disproportionate number of our contributors (and current employees)
come from the Linux user community. About 25% more people use Nightly on
Linux than OSX.

- mhoye
_______________________________________________
dev-planning mailing list
[hidden email]
https://lists.mozilla.org/listinfo/dev-planning
Reply | Threaded
Open this post in threaded view
|

Re: The future of Linux32 builds and tests

Chris Coulson-2
In reply to this post by Daniel Holbert-3
On 09/10/15 02:01, Daniel Holbert wrote:

> On 10/08/2015 02:43 PM, William Lachance wrote:
>> Aren't the vast majority of users running a version of Firefox
>> distributed by their vendor (of which Ubuntu probably has the largest
>> share, see e.g.
>> http://stats.wikimedia.org/wikimedia/squids/SquidReportOperatingSystems.htm)?
>> I don't think any metrics gathered from the Firefox builds which we
>> distribute are going to be very useful for measuring the impact of this
>> change.
> Oh, right -- it looks like telemetry isn't enabled in my Ubuntu-provided
> Firefox build.  (The telemetry checkbox shown in the "how to enable
> telemetry" article[1] just isn't present.)  I assumed it would be
> enabled (and opt-in for users), but I assumed wrong.
>
> So, it seems we aren't getting telemetry from Ubuntu users at least. So,
> unless we're getting platform info some other way (AUS pings?), sounds
> like we can't be sure about 32-vs-64 market share without talking to
> distros.
>
> ~Dnaiel
>
> [1]
> https://support.mozilla.org/en-US/kb/share-telemetry-data-mozilla-help-improve-firefox
> _______________________________________________
> dev-planning mailing list
> [hidden email]
> https://lists.mozilla.org/listinfo/dev-planning
Hi,

I'm not sure I can provide any better metrics. You get blocklist pings
from us though, don't you? Do these provide a useful metric?

Regards
- Chris
_______________________________________________
dev-planning mailing list
[hidden email]
https://lists.mozilla.org/listinfo/dev-planning
Reply | Threaded
Open this post in threaded view
|

Re: The future of Linux32 builds and tests

Robert Kaiser
In reply to this post by danderson
[hidden email] schrieb:
> According to Telemetry, Linux has about 0.2% market share on Desktop, which means Windows has around 500X more users.

That number does not look right. Our ADI numbers say that 3.5% of our
overall ADI are on Linux. Given that a number of Linux distros may
switch off Telemetry and FHR by default, but leave the add-on blocklist
enabled (where ADI comes from), I'd trust that number some more - even
with the flaws that the ADI measurement might have.

That said, the fact that Linux distros might switch off Telemetry and
FHR by default causes issues right in that way: they are undercounted
and we may disregard them completely in data-driven decisions. If you
are a Linux distro package manager, think about that.

KaiRo

_______________________________________________
dev-planning mailing list
[hidden email]
https://lists.mozilla.org/listinfo/dev-planning
Reply | Threaded
Open this post in threaded view
|

Re: The future of Linux32 builds and tests

Robert Kaiser
In reply to this post by Lawrence Mandel-2
Mike Hommey schrieb:
> Debian is likely not very indicative of the overall Linux world. Taking
> the primarily available figures from Debian and Ubuntu give completely
> opposite pictures:
> debian: i386: 51859, amd64: 132362
> ubuntu: i386: 2065695, amd64: 673565

And even other distros may give yet different numbers. For example, for
the openSUSE Leap distro, 32bit will not even be an officially supported
architecture any more, only x86_64 will be. Unfortunately, I have no
clue if Fedora, RedHat, openSUSE, Arch and others even do have numbers
like you point out above.

KaiRo

_______________________________________________
dev-planning mailing list
[hidden email]
https://lists.mozilla.org/listinfo/dev-planning
Reply | Threaded
Open this post in threaded view
|

Re: The future of Linux32 builds and tests

Mike Connor-4
I think ADI is likely the only reliable metric here, if FHR and Telemetry
are disabled by Linux distributions. That said, if folks have examples of
Linux distros explicitly disabling these features, please let me know.
Unless they've got written permission to disable those features, they
shouldn't be doing that.  And ultimately it's self-defeating, as it means
Linux users (and their use cases) won't be measured or prioritized
effectively.

-- Mike

On 9 October 2015 at 09:23, Robert Kaiser <[hidden email]> wrote:

> Mike Hommey schrieb:
>
>> Debian is likely not very indicative of the overall Linux world. Taking
>> the primarily available figures from Debian and Ubuntu give completely
>> opposite pictures:
>> debian: i386: 51859, amd64: 132362
>> ubuntu: i386: 2065695, amd64: 673565
>>
>
> And even other distros may give yet different numbers. For example, for
> the openSUSE Leap distro, 32bit will not even be an officially supported
> architecture any more, only x86_64 will be. Unfortunately, I have no clue
> if Fedora, RedHat, openSUSE, Arch and others even do have numbers like you
> point out above.
>
> KaiRo
>
>
> _______________________________________________
> dev-planning mailing list
> [hidden email]
> https://lists.mozilla.org/listinfo/dev-planning
>
_______________________________________________
dev-planning mailing list
[hidden email]
https://lists.mozilla.org/listinfo/dev-planning
Reply | Threaded
Open this post in threaded view
|

Re: The future of Linux32 builds and tests

William Lachance-2
In reply to this post by Lawrence Mandel-2
On 15-10-08 09:40 PM, Mike Hommey wrote:
> Debian is likely not very indicative of the overall Linux world. Taking
> the primarily available figures from Debian and Ubuntu give completely
> opposite pictures:
> debian: i386: 51859, amd64: 132362
> ubuntu: i386: 2065695, amd64: 673565
>
> (from popcon.ubuntu.com and popcon.debian.org)

Are those Ubuntu numbers for installations in general, or Firefox
specifically? I couldn't find out how to get architecture specific
information from popcon, and I'm worried that the overall results
include server installations (which would really skew them):

http://popcon.ubuntu.com/main/index.html

If that was referring to Firefox, it's probably representative. It
appears popularity contest is installed on every Ubuntu instance:

https://help.ubuntu.com/community/UbuntuPopularityContest

Even without specific numbers, the fact that Ubuntu continues to see a
32-bit desktop version of their desktop operating system as worth
shipping and supportive is probably indicative in itself.

Will
_______________________________________________
dev-planning mailing list
[hidden email]
https://lists.mozilla.org/listinfo/dev-planning
12