[webkit-qt] QtWebkit2.1 version number

Simon Hausmann simon.hausmann at nokia.com
Tue Dec 21 06:35:59 PST 2010

On Tuesday, December 21, 2010 09:37:26 am ext Koskinen Janne wrote:
> Hi all, this is my last attempt on this even if the bug was closed etc.
> QtWebkit.sis on QtWebkit2.1 branch is currently versioned as 4.8.0.
> If we keep this version when Qt 4.8 is out or new version of QtWebkit is
> released it will need to be incremented. This means that Qt 4.8 would have
> QtWebkit version 4.9 or that Qt 4.8 would have to ship with QtWebkit2.1.
> Problem is that user will see this number if manually installing. Installer
> prompts "xxx needs QtWebkit 4.8.0(0), are you sure you want to install?"
> and will be pretty bad as there is no such version anywhere.

IMHO this is not a technical problem but an issue with the UI of the Symbian 
installer, that in the first place shouldn't present the users with questions 
about version numbers users shouldn't have to know about in the first place.

I mean seriously, based on the provided information ("app needs foo 4.8.0"), 
how can the user actually make an informed decision? ("are you sure?")

So IMHO we shouldn't change version numbers just to work around UI issues in 
> Worse problem is that smart installer will use these numbers to determine
> which version of QtWebkit is required to run application x. We cannot
> change the sis file numbers to 2.1.0 as we have already deployed Qt and
> there are applications that depend on QtWebkit version 4.6.3 or 4.7.0 and
> as such those version would have to be installed to be able to run them as
> they both are higher version than 2.x.x.

Right, so if we don't change anything right now, stick to 4.8, 4.9, etc. , 
we'll be fine, no?
> This is simplification of the issue, add Symbian file eclipising, add
> version number size restrictions and you start to feel why this is baaad.

I admit I don't see why 4.8 and 4.9, etc. are bad. What other issues are 
> My original suggestion was to have QtWebkit 2.1.0 sis version to be
> 20.1.0(0). Issue won't manifest until it is too late to do anything about
> it.

If there is a real technical problem, then I'm all for investigating a binary 
version change. If the problem is Symbian specific, then maybe we should look 
for a symbian / sis-file specific solution (different versioning there?). If the 
problem can be limited to "confusing things presented to the user", then I 
think we should ignore the problem :)


More information about the webkit-qt mailing list