Merging the android plugin to Qt Creator
mingw.android at gmail.com
Thu Sep 1 20:37:23 UTC 2011
Many thanks Daniel.
I applied and pushed these commits to 2.3-stagging.
On Thu, Sep 1, 2011 at 7:24 PM, Daniel Teske <daniel.teske at nokia.com> wrote:
> On Thursday 01 Sep 2011 15:43:05 Teske Daniel (Nokia-MP-Qt/Berlin) wrote:
> > > I tried to apply your patches 2.3-stagin but it doesn't compile
> > > should I create another branch and merge master branch of upstream
> > > qt-creator ? This mean to create another merge request !
> > The patches were on top of a master merge into 2.3-staging.
> > So 0002 and 0003 should apply cleanly, whereas you 0001 is only needed if
> > you want to merge master into 2.3-staging. That depends on your plans for
> > the next sdk release. I'd guess you would probably want to use 2.3 for
> > now, which means, you should save 0001 for later.
> I did recreate those patches against 2.3-staging and added a few more:
> 0001 and 0002 are just the rebased patches.
> I did a first pass over the android plugin code and fixed all coding style
> errors I found. Please check this patch carefully, I'm pretty sure that
> the amount of small changes, I might have made a conversion error
> The major ones were:
> - put a space after "if", "while", "for" and others.
> - put spaces around operators like "+", "-", "+="
> - put a space after "," and ";" in the for loop
> - use isEmpty() instead if size() or lenght() (for no particular reason
> - "Type &variable" is our prefered way
> - braces (except for classes and functions) don't have a line for
> - the branches in if/else, either both or none should have braces
> Various small things
> I also probably missed quite a few cases, but that's a start. :)
> For some reason, without that patch the branch doesn't compile for me
> QLatin1String is essentially just a const char *, thus passing a reference
> it, is the same as passing a reference to a pointer, that's not faster then
> passing the pointer directly.
> I think the QProcess is leaked if the process could not be started for any
> reason, since in that case no finished signal is emitted.
> The using isEnabled() function was to work around that virtual functions
> hidden if they are overloaded in a derived class with a different parameter
> list. I removed that problem before 2.3, there's now no overloading
> thus the using line is simply no longer needed
> The cast is not needed if we pass on the right type
> The class doesn't exist anymore in that namespace, and indeed it wasn't
> The code looked wrong, though I'm not sure whether that fix is the correct
> I haven't actually tested it.
> The top level qtcreator.pro checks whether qt is at least 4.7.4, thus the
> can be savely removed
> We don't use abbreviations in our names, thus "Dont" isn't allowed, also
> NoMatch seems to be a better fit. The other changes should be pretty clear.
> Necessitas-devel mailing list
> Necessitas-devel at kde.org
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Necessitas-devel