Closed
Bug 844053
Opened 12 years ago
Closed 12 years ago
Separate win8 updates from other windows updates
Categories
(Release Engineering :: Release Requests, defect)
Release Engineering
Release Requests
Tracking
(Not tracked)
VERIFIED
FIXED
People
(Reporter: catlee, Unassigned)
References
Details
Attachments
(1 file, 2 obsolete files)
1.05 KB,
patch
|
catlee
:
review+
akeybl
:
feedback+
bhearsum
:
checked-in+
|
Details | Diff | Splinter Review |
We may be required to provide updates to 19.0.1 to *only* win8 users. Currently AUS doesn't support this kind of restriction, so some hacks are required.
Here's the best approach I can come up with. If we see a win8 user for Firefox (identified by 'Windows_NT 6.2' in the OS_VERSION string), then we change what AUS platform we consider them to be. We can then publish snippets that target only win8 users.
For 19.0.1, this could mean that we change the update_platform in automation so that snippets for windows builds get created with the right name. Or we could rename the snippet directory after they're generated.
Does this make any sense?
Can somebody please double check that 'Windows_NT 6.2' is the right string to be looking for?
If we can find a better approach, then great. However, the first thing to ask yourself if you think you have a better idea is: does it update a win7 18.0 user to 19.0 while at the same time updating a win8 18.0 user to 19.0.1?
This patch will have to be removed before 19.0.2 / 20.0. Perhaps it should be limiting its behaviour to just release/releasetest channels as well?
Attachment #717063 -
Flags: feedback?(rhelmer)
Attachment #717063 -
Flags: feedback?(rail)
Reporter | ||
Comment 1•12 years ago
|
||
Same as before, except limit to release, releasetest channels only
Attachment #717063 -
Attachment is obsolete: true
Attachment #717063 -
Flags: feedback?(rhelmer)
Attachment #717063 -
Flags: feedback?(rail)
Attachment #717119 -
Flags: feedback?(rhelmer)
Attachment #717119 -
Flags: feedback?(rail)
Comment 2•12 years ago
|
||
Comment on attachment 717063 [details] [diff] [review]
Change aus platform for win8 users
Review of attachment 717063 [details] [diff] [review]:
-----------------------------------------------------------------
::: index.php
@@ +105,5 @@
> +// To accomplish this, we change the platform of win8 users, so we can look in
> +// a distinct snippet directory for them.
> +if ($clean['product'] == 'Firefox' &&
> + $clean['platform'] == 'WINNT_x86-msvc' &&
> + strpos($clean['platformVersion'], 'Windows_NT 6.2')) {
strpos($clean['platformVersion'], 'Windows_NT 6.2')) === true
probably would be better here, otherwise lgtm.
I checked the update URL on a W8 machine and it was like ..."Windows_NT%206.2.0.0"...
Attachment #717063 -
Attachment is obsolete: false
Updated•12 years ago
|
Attachment #717119 -
Flags: feedback?(rail) → feedback+
Updated•12 years ago
|
tracking-firefox19:
--- → ?
Comment 3•12 years ago
|
||
Would this patch also help us to disable updates to FF19.0 for Win8 users only on the Release channel?
Updated•12 years ago
|
Attachment #717119 -
Flags: feedback?(rhelmer) → feedback+
Comment 4•12 years ago
|
||
(In reply to Alex Keybl [:akeybl] from comment #3)
> Would this patch also help us to disable updates to FF19.0 for Win8 users
> only on the Release channel?
I believe so, yes. If I understand it correctly, it should give us the ability to serve unique snippets to all combinations of win8 + version.
Comment 5•12 years ago
|
||
(In reply to Rail Aliiev [:rail] from comment #2)
> Comment on attachment 717063 [details] [diff] [review]
> Change aus platform for win8 users
>
> Review of attachment 717063 [details] [diff] [review]:
> -----------------------------------------------------------------
>
> ::: index.php
> @@ +105,5 @@
> > +// To accomplish this, we change the platform of win8 users, so we can look in
> > +// a distinct snippet directory for them.
> > +if ($clean['product'] == 'Firefox' &&
> > + $clean['platform'] == 'WINNT_x86-msvc' &&
> > + strpos($clean['platformVersion'], 'Windows_NT 6.2')) {
>
> strpos($clean['platformVersion'], 'Windows_NT 6.2')) === true
> probably would be better here, otherwise lgtm.
>
> I checked the update URL on a W8 machine and it was like
> ..."Windows_NT%206.2.0.0"...
Where's you check this? I'd like a full update URL example so we can do some more testing.
Comment 6•12 years ago
|
||
I found a Windows 8 machine. Here's a full update URL for Firefox 19.0:
https://aus3.mozilla.org/update/3/Firefox/19.0/20130215130331/WINNT_x86-msvc/en-US/release/Windows_NT%206.2.0.0%20(x64)/default/default/update.xml?force=1
Comment 7•12 years ago
|
||
And I've got this page setup on aus3-staging, the URL I'm using to test so far is: http://10.8.74.30/update/3/Firefox/18.0.2/20130201065344/WINNT_x86-msvc/en-US/release/Windows_NT%206.2.0.0%20%28x64%29/default/default/update.xml?force=1.
Comment 8•12 years ago
|
||
Initial testing showed the previous patch not to work. This version fixes the platformVersion condition to work. Still more testing to do, just wanted to upload this before I forgot.
Attachment #717063 -
Attachment is obsolete: true
Attachment #717119 -
Attachment is obsolete: true
Comment 9•12 years ago
|
||
Comment on attachment 718423 [details] [diff] [review]
fix platformVersion condition
OK. I think this is working fine. I setup some test snippets on the "bhearsumtest" channel that:
* Point 18.0.2 non-win8 @ 19.0 (same as what's currently on "release")
* Point 19.0 win8 @ a fake 20.0
Here's non-win8 18.0.2: http://10.8.74.30/update/3/Firefox/18.0.2/20130201065344/WINNT_x86-msvc/en-US/bhearsumtest/Windows_NT%206.1.0.0%20%28x64%29/default/default/update.xml?force=1
Here's non-win8 19.0: http://10.8.74.30/update/3/Firefox/19.0/20130226999999/WINNT_x86-msvc/en-US/bhearsumtest/Windows_NT%206.1.0.0%20%28x64%29/default/default/update.xml?force=1
Here's win8 19.0: http://10.8.74.30/update/3/Firefox/19.0/20130226999999/WINNT_x86-msvc/en-US/bhearsumtest/Windows_NT%206.2.0.0%20%28x64%29/default/default/update.xml?force=1
In the real release we can add snippets for 18.0.2 and earlier win8 too, I didn't bother with that for testing purposes though.
Since we're already in a state where Windows updates are throttled (instead of blocked via unsupportedPlatforms), nightly builds are already getting updates. Eg: https://aus3.mozilla.org/update/3/Firefox/13.0a1/20120222174716/WINNT_x86_64-msvc/en-US/nightly/Windows_NT%206.2.1.0%20%28x64%29/default/default/update.xml?force=1 - so I think that addresses Alex's concern from comment #3.
Attachment #718423 -
Flags: review?(catlee)
Attachment #718423 -
Flags: feedback?(akeybl)
Reporter | ||
Updated•12 years ago
|
Attachment #718423 -
Flags: review?(catlee) → review+
Comment 10•12 years ago
|
||
Comment on attachment 718423 [details] [diff] [review]
fix platformVersion condition
From the explanation given, this is what we want. For the actual push, I filed bug 845447 (utilizing this patch).
Attachment #718423 -
Flags: feedback?(akeybl) → feedback+
Comment 11•12 years ago
|
||
(In reply to Ben Hearsum [:bhearsum] from comment #9)
> OK. I think this is working fine. I setup some test snippets on the
> "bhearsumtest" channel that:
> * Point 18.0.2 non-win8 @ 19.0 (same as what's currently on "release")
> * Point 19.0 win8 @ a fake 20.0
Anthony - could you help confirm Win8 vs non-Win8 findings on bhearsumtest? Using needinfo instead of qacontact due to conflict.
Flags: needinfo?(anthony.s.hughes)
Keywords: qawanted
Comment 12•12 years ago
|
||
(In reply to Alex Keybl [:akeybl] from comment #11)
> (In reply to Ben Hearsum [:bhearsum] from comment #9)
> > OK. I think this is working fine. I setup some test snippets on the
> > "bhearsumtest" channel that:
> > * Point 18.0.2 non-win8 @ 19.0 (same as what's currently on "release")
> > * Point 19.0 win8 @ a fake 20.0
>
> Anthony - could you help confirm Win8 vs non-Win8 findings on bhearsumtest?
> Using needinfo instead of qacontact due to conflict.
Argh...I just deleted these snippets. I can recreate them if necessary.
Comment 13•12 years ago
|
||
(In reply to Ben Hearsum [:bhearsum] from comment #12)
> (In reply to Alex Keybl [:akeybl] from comment #11)
> > (In reply to Ben Hearsum [:bhearsum] from comment #9)
> > > OK. I think this is working fine. I setup some test snippets on the
> > > "bhearsumtest" channel that:
> > > * Point 18.0.2 non-win8 @ 19.0 (same as what's currently on "release")
> > > * Point 19.0 win8 @ a fake 20.0
> >
> > Anthony - could you help confirm Win8 vs non-Win8 findings on bhearsumtest?
> > Using needinfo instead of qacontact due to conflict.
>
> Argh...I just deleted these snippets. I can recreate them if necessary.
Though, I don't know if QA has access to aus3-staging.mozilla.org to test.
Comment 14•12 years ago
|
||
We can always go with bug 845447 once staged, just wanted to get early feedback.
Comment 15•12 years ago
|
||
Comment on attachment 718423 [details] [diff] [review]
fix platformVersion condition
Checking in index.php;
/cvsroot/mozilla/webtools/aus/xml/index.php,v <-- index.php
new revision: 1.38; previous revision: 1.37
done
Attachment #718423 -
Flags: checked-in+
Reporter | ||
Updated•12 years ago
|
Status: NEW → RESOLVED
Closed: 12 years ago
Resolution: --- → FIXED
Summary: Update only win8 users to 19.0.1 → Separate win8 updates from other windows updates
Updated•12 years ago
|
tracking-firefox19:
? → ---
Flags: needinfo?(anthony.s.hughes)
Comment 16•12 years ago
|
||
This appears to be working as expected. Windows 8 users (and only Windows 8 users) are receiving updates to 19.0.1 on release channel. All other users are receiving Firefox 19.0.
Comment 18•12 years ago
|
||
Note: if we do this again we should include 'betatest' in the channel list.
Assignee | ||
Updated•11 years ago
|
Product: mozilla.org → Release Engineering
You need to log in
before you can comment on or make changes to this bug.
Description
•