Bug 38389 – [Qt] Update the Symbian version for the user agent

• Status: CLOSED
• Resolution: FIXED
• Component: New Bugs

[Benjamin Poulain]

The current user agent string does not take into account the new versions of Symbian:
Symbian^2, Symbian^3 and Symbian^4.

[Benjamin Poulain]

Created attachment 54914 [details]
Patch with SymbianOS^2 to SymbianOS^4

[Benjamin Poulain]

Created attachment 54915 [details]
Alternative: use SymbianOS/XX for the old scheme, and Symbian^X for the new scheme

[Benjamin Poulain]

Note that even if the second patch is preferred, the string SymbianOS will still be in the user agent because the string currently use "SymbianOS" as platform.

[Laszlo Gombos]

As SV_SF_1 is the same than SV_9_4, we decide if it make sense to change 'SymbianOS/9.4" to "Symbian^1". At the moment it is safer to keep it as it is, but I'd like to at least consider it.

[Benjamin Poulain]

(In reply to comment #4)
> As SV_SF_1 is the same than SV_9_4, we decide if it make sense to change
> 'SymbianOS/9.4" to "Symbian^1". At the moment it is safer to keep it as it is,
> but I'd like to at least consider it.

Is there any Symbian^1 device?
I thought it was S60v5 re-branded for historic reason.

[Jesus Sanchez-Palencia]

I don't think this is QuickTime related; changing to Qt. ;-)

[Kenneth Rohde Christiansen]

(In reply to comment #4)
> As SV_SF_1 is the same than SV_9_4, we decide if it make sense to change
> 'SymbianOS/9.4" to "Symbian^1". At the moment it is safer to keep it as it is,
> but I'd like to at least consider it.

Why safer? :-) If people are testing for SymbianOS today, it would be nice if their tests fail so that we don't get the somewhat crappy Symbian optimized pages.

[Kenneth Rohde Christiansen]

(In reply to comment #5)
> (In reply to comment #4)
> > As SV_SF_1 is the same than SV_9_4, we decide if it make sense to change
> > 'SymbianOS/9.4" to "Symbian^1". At the moment it is safer to keep it as it is,
> > but I'd like to at least consider it.
> 
> Is there any Symbian^1 device?
> I thought it was S60v5 re-branded for historic reason.

Exactly, ^1 is rebranding, ^2 is opensource version of ^1, ...

[Kenneth Rohde Christiansen]

Comment on attachment 54915 [details]
Alternative: use SymbianOS/XX for the old scheme, and Symbian^X for the new scheme

So now we have:

SymbianOS/XX
Symbian^X
Symbian/XX (Unknown)

Will the Symbian^2 etc have subversions in the future? like Symbian^2/9.2 or anything like that?

How should we treat the unknown? I just using just Symbian or SymbianOS would be fine.

[Benjamin Poulain]

(In reply to comment #9)
> Symbian/XX (Unknown)
> 
> How should we treat the unknown? I just using just Symbian or SymbianOS would
> be fine.

I kept the "/" because I don't know if Unknown can refer to and older Symbian, or a Symbian running on a non-Nokia phone.

[Kenneth Rohde Christiansen]

(In reply to comment #10)
> (In reply to comment #9)
> > Symbian/XX (Unknown)
> > 
> > How should we treat the unknown? I just using just Symbian or SymbianOS would
> > be fine.
> 
> I kept the "/" because I don't know if Unknown can refer to and older Symbian,
> or a Symbian running on a non-Nokia phone.

I don't think anyone would actually look for "unknown" :-) I guess you handle what you can handle and then just use a fallback for the rest.

Maybe Laszlo has some insight, or Zalan?

[Laszlo Gombos]

(In reply to comment #7)
> (In reply to comment #4)
> > As SV_SF_1 is the same than SV_9_4, we decide if it make sense to change
> > 'SymbianOS/9.4" to "Symbian^1". At the moment it is safer to keep it as it is,
> > but I'd like to at least consider it.
> 
> Why safer? :-) If people are testing for SymbianOS today, it would be nice if
> their tests fail so that we don't get the somewhat crappy Symbian optimized
> pages.

Yes, I had the same reasoning, but I have no hands on data if this is really the case.

[Benjamin Poulain]

Comment on attachment 54915 [details]
Alternative: use SymbianOS/XX for the old scheme, and Symbian^X for the new scheme

Test Bugzilla

[Benjamin Poulain]

Comment on attachment 54915 [details]
Alternative: use SymbianOS/XX for the old scheme, and Symbian^X for the new scheme

Restore flags

[Laszlo Gombos]

Comment on attachment 54914 [details]
Patch with SymbianOS^2 to SymbianOS^4

Looks good to me, thanks. r+.

[Laszlo Gombos]

Comment on attachment 54915 [details]
Alternative: use SymbianOS/XX for the old scheme, and Symbian^X for the new scheme

r- as we decided to go with the first patch only. 

If we find an actual (maybe top 100 or so) case where changing the SymbianOS make a difference, we should reconsider this approach.

[WebKit Commit Bot]

Comment on attachment 54914 [details]
Patch with SymbianOS^2 to SymbianOS^4

Clearing flags on attachment: 54914

Committed r59451: <http://trac.webkit.org/changeset/59451>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.

[Simon Hausmann]

Revision r59451 cherry-picked into qtwebkit-2.0 with commit 7eb1783afff639ecc1e0b57eb5f341171dd99afd

[Laszlo Gombos]

Here is the latest proposal for the "final" UA string for N8. The closest QtWebKit can get to the following UA string the more likely we will avoid fragmentation between QtWebKit clients.

"Mozilla/5.0 (Symbian/3; U; NokiaN8-00; Profile/MIDP-2.1 Configuration/CLDC-1.1
en-US) AppleWebKit/532.4 (KHTML, like Gecko) QtLauncher/1.0 Mobile Safari/532.4

[Laszlo Gombos]

Reopen to fix the UA for Symbian port of QtWebKit as suggested above and fix regression introduced by r58648.

[Laszlo Gombos]

Created attachment 56547 [details]
Symbian (and Maemo5) UA fixes

[Laszlo Gombos]

Created attachment 56550 [details]
patch for 4.6 (untested)

[Laszlo Gombos]

Tested patch to fill out the model number and CLDC_MIDP version (e.g. "NokiaN8-00; Profile/MIDP-2.1 Configuration/CLDC-1.1") would be welcome - either using public Symbian APIs or perhaps QtMobility APIs (SysInfo?).

[Laszlo Gombos]

Created attachment 56553 [details]
patch for 4.6 v2 - add "Mobile" to the UA string as well

[Benjamin Poulain]

Do you really want SymbianOS/4 for Symbian^4? What happen if Symbian numbering goes to 9: Symbian^9 != SymbianOS/9.

[Laszlo Gombos]

I think the attached patch makes the version "Symbian/4" and not "SymbianOS/4" for Symbian^4.

The idea is that we use "SymbianOS" for the "old" Symbian numbering and "only" "Symbian" for the "new" Symbian numbering.

I like this idea as it makes the UA overall shorter going forward - and at the same time it is clear which generation of version UA refers to.

This is also inline with the actual re-scoping of Symbian Foundation as Symbian now includes more SW components than just lower "OS level" pieces.

[Benjamin Poulain]

(In reply to comment #27)
> I think the attached patch makes the version "Symbian/4" and not "SymbianOS/4" for Symbian^4.

You are right. I misunderstood the intent.

[Kenneth Rohde Christiansen]

Comment on attachment 56547 [details]
Symbian (and Maemo5) UA fixes

Fine with me. Btw, should we change the Q_WS_MAEMO_5 test to MAEMO? is MAEMO still defined on Maemo 6 and beyond? We will also soon have to deal with MeeGo, but that is more complicated as MeeGo is not only for mobile devices.

[WebKit Commit Bot]

Comment on attachment 56547 [details]
Symbian (and Maemo5) UA fixes

Clearing flags on attachment: 56547

Committed r60030: <http://trac.webkit.org/changeset/60030>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.

[Simon Hausmann]

Revision r60030 cherry-picked into qtwebkit-2.0 with commit 815dbb2d4ab288a51e966fac798f852494bd8b5d

[Laszlo Gombos]

Created attachment 57404 [details]
patch for 4.6 v3 (created with git diff)

[Simon Hausmann]

(In reply to comment #33)
> Created an attachment (id=57404) [details]
> patch for 4.6 v3 (created with git diff)

Cherry-picked into qtwebkit-4.6 with commit 3bb0df4f00fa0b3ac3abc1b6fe508080fabdeaab