[Qgis-developer] Broken code after upgrade to QGIS 1.4

[Matthew Denno]

Martin,

Thanks for replying to my email.

I have done a bit more looking at my problem and it looks like the
problem really lies (maybe) in the qatomic_i386.h file with the inline
bool QBasicAtomicInt::ref() function.  If I don't step into the code I
sent before that is where it fails.  I am using the Qt Creator IDE,
which I like but seem to have trouble debugging with.

I have answered your questions below, but am not sure if maybe my
recent findings change things.  I need to do some more Google-ing
probably.

On Tue, Jan 19, 2010 at 7:03 AM, Martin Dobias <wonder.sk at gmail.com> wrote:
> Hi Matthew
>
> On Tue, Jan 19, 2010 at 5:16 AM, Matthew Denno <mgdenno at gmail.com> wrote:
>> Hello List,
>>
>> Quick question.  The following piece of code seemed to always work
>> before upgrading to 1.4, but now it seems to fail with this message:
>>
>> The program has unexpectedly finished.
>> /home/matt/Programming/DelaunayIntersection/DelaunayIntersection
>> exited with code 0
>>
>> Any ideas with what the cause might be?  Any help or direction would
>> be greatly appreciated.
>
> Just making sure: have you recompiled your code after upgrade to 1.4?

I am new to C++ so that is a good question, but yes I have recompiled
my code and it compiled fine.

> Is QGIS 1.4 itself working fine? (e.g. it opens a shapefile without problems)

Yes, seems to work fine.

> Is that a standalone application or a plugin?

It is a stand alone application.

> If it's an application, do you initialize QgsApplication?

I don't initialize QgsApplication.  Is this required?  I didn't do
this previously either.  My application performs an analysis and
generates a few shapefiles but does not display them.  On a side note
I would like to make it a plugin in the future, but am working out
some other issues first.

> Is the provider path in your code (/usr/lib/qgis) still valid?

Yes it seems to be.

>
> The code itself looks fine and shouldn't be affected by upgrades.
> (Except for that you create an instance of QgsVectorLayer and never
> free it.)

Oops! :-)

>
> Martin
>