[Kmymoney-devel] Some planning for the next release

[Alvaro Soliverez]

On Fri, Feb 5, 2010 at 11:17 AM, Fernando Vilas <fvilas at iname.com> wrote:
> On Friday 05 February 2010 06:41:32 Alvaro Soliverez wrote:

>> I want to focus my work on making it easier to integrate with other
>> apps. As part of that, some code has to be taken out of kmymoney.cpp
>> and into some deeper class.
>> There are several reasons for this:
>> - The logic has to be split from the GUI elements. When trying to
>> write an engine for Akonadi, this has proven to be a problem.
>> - Kmymoney.cpp is upwards from 6000+ LOC and growing, way beyond what
>> is manageable.
>> - The code for loading and saving the file is in there. As it is
>> attached to GUI code, it can't be unit tested, which has been a major
>> pain in the past and will be again in the future, for sure.
>> - Other applications can't easily use kmymoney libs unless they
>> rewrite the loading/saving code
>>
>> If this is done, at least for the major methods, it should be done
>> during the next release or wait until after stable release, because it
>> has to be well tested.
>>
>
> If this can be done in small increments, I'm fine with having it in the next
> release. Otherwise, I vote for waiting until after the stable release.
>

Yes, it will be done at a conservative pace, as we have done so far,
and which is paying off in good reviews even when still in beta, I
think.

What I intend is to plant the seed for a pretty major change after
stable, but that seed would still allow us to do some pretty
interesting and yet it's small enough to test it thoroughly during one
minor cycle.

> After stable release, I want to attempt to remove the dependency on CPPUNIT
> and use Qt for the unit tests. This provides the advantage of being able to
> unit test gui classes by simulating button clicks, etc. so the event handlers
> get called in context. It is also something that touches too many files when we
> are in beta and trying squash bugs before a stable release.
>

That will be a major undertaking, with almost 400 tests out there.

> I'm also still working the db speed issue by consolidating some calls to
> reduce round trips. I'm trying to make sure that it is also very stable before
> I commit, otherwise that patch will also wait until after stable release. This
> is something small enough that it may be able to go in another minor release
> before stable.
>

We expect the change in models to help with that, because it will
require fewer loads to display the same data throughout the
application. The database will get special benefits out of this, I
hope.

Regards,
Alvaro