Merging the android plugin to Qt Creator

[Daniel Teske]

Hi,

I'm back from vacation and had another look at the diff between master and 2.3-
staging.[1] For that I merged master into 2.3-staging, since a few changes 
that I did in master aren't yet in the 2.3-staging branch.

There are currently 5 merge conflicts, none particular complicated to figure 
out. Unfourtanetly that didn't compile, due to a header cleanup patch by 
Friedemann. For reference, the 0001-Compile-fixes is what was required to make 
it compile again. 
Also the exporting FolderNavigationWidget is no longer needed, since the 
showInGraphicalShell method is moved in master, see 0001-Compile-fixes

Also I have prepared two patches for the android branch:
- 0002-Remove-no-longer-used-ressource.h 
The file is a left-over from the fix Ray did to the qtcreator.rc file. Since it's 
now fixed in a different way, the file is no longer needed.

- 0003-Code-stylistics-to-RemoteGdbServerAdapter-changes.patch
I have adjusted the code a little bit to fit the creator code sytle a little 
better. That is using !isEmpty() instead of size() and adding spaces around 
operators. Andre thinks the patch is now fine.

Also
> n) lowerCasing of paths in qt4buildconfiguration.cpp
That turns out to be somewhat more involved, I still have it on my todo list.

> j) bin/necessitas and bin/necessitas.bat
Ossi reintroduced the LD_LIBRARY_SCRIPT, I still don't really understand the 
topic.

> s) main.cpp #ifdef
> There is a change in main.cpp changing the plugin search paths to be only
> searched on the "matching" platform.
I cherry-picked that change to master.

> p) debugger plugin
> Disabling setting PYTHONPATH. Why is that needed? Note that the code that
> is disabled was rewritten a lot since you disabled it.
We removed the same code that was #ifdefed out in master now, so on merging, 
just remove the code. 

So going over my intial todo list, I still have a few open questions. As far 
as I can tell they are all for Ray. :)

> m) qmakeStep::moreArguments()
> Adding the "-win32" to the qmake command line. As far as I understand you
> want to have qmake run in HOST_WIN_MODE on windows. But that is the
> default if qmake is compiled for windows.
 
> p) debugger plugin
> Always offering all toolchains in DebuggerToolChainComboBox::init
> The commit message already classifies that as a hack, I just don't
> understand why it was needed. The check for hostAbi.os() == abi.os()
> should be true for msys toolchains too. Or is .os() broken for msys
> toolchains?

and
> t) main.cpp myMessageOutput
> The main motivation for that seems to be:
> "qInstallMsgHandler so that asserts are not fatal"
> We like Q_ASSERTS to be fatal, we do have a separate macro QTC_ASSERT for
> cases where a non fatal handling is possible/desired.

Ray, any comments on them would be nice.

 
daniel

[1] That seemed to be the most recent branch
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0003-Code-stylistics-to-RemoteGdbServerAdapter-changes.patch
Type: text/x-patch
Size: 1513 bytes
Desc: not available
URL: <http://mail.kde.org/pipermail/necessitas-devel/attachments/20110831/59dbed36/attachment.patch>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0002-Remove-no-longer-used-ressource.h.patch
Type: text/x-patch
Size: 975 bytes
Desc: not available
URL: <http://mail.kde.org/pipermail/necessitas-devel/attachments/20110831/59dbed36/attachment-0001.patch>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0001-Compile-fixes.patch
Type: text/x-patch
Size: 11630 bytes
Desc: not available
URL: <http://mail.kde.org/pipermail/necessitas-devel/attachments/20110831/59dbed36/attachment-0002.patch>