[Development] Qwt under the qt-project umbrella
Uwe.Rathmann at tigertal.de
Sun Apr 15 19:05:45 CEST 2012
Hi Lars + Thiago,
let me continue in a new thread:
On 04/15/2012 12:58 PM, lars.knoll at nokia.com wrote:
> Look at the definition of a Qt add-on. I think that might fit nicely
> In any case, if you want your library do live under the qt-project
> umbrella we can most likely find a solution.
I didn't get all details of the rules, but my feeling tells me that the
spirit of qt-project makes it the natural home for a project like Qwt.
But in the end the answer to Thiagos question "why do you want to?" counts:
Today Qwt is hosted at sourceforge offering me:
- a project web page
- a mailing list
- a bug and task tracker ( I don't use much )
- a service how to release software
- a donation ssystem
I had almost no administration overhead with it since more than 10
years, what is something.
But there are significant problems I can imagine to be less when moving
1) Binary packages
I guess about 50% of the support requests are about how to build Qwt to
work with the Qt SDK on windows.
Especially getting the Qwt designer plugin loaded into the creator seems
to be a nightmare when you are not familiar with Qt.
Today I run a couple example applications on my Linux/X11 box only - I
don't even have a Windows system. I'm surprised myself, that this works
Being part of some automatic cross platform regression tests would be a
The documentation is by far the weakest part of Qwt.
I don't expect someone to do the documentation work for me, but some
support would be great.
4) Better communication
Even if Qwt is the oldest and most popular ( beside KDE of course ) Qt
addon I guess none of the Qt core developers knows much more about it
than that it exists.
Finding someone to do an API ( or better code ) review only - like
Thiago suggested - would improve a lot.
After explaining my general expectations let me answer to Thiagos comments:
> I'd prefer to see a full API review, just as what we did for some
> that moved into Qt ...
The headers of Qwt alone sum up to more than 12k lines of code !
If one wants to do it with a certain level of quality ( not only
checking formal stuff ) it means to understand some concepts and what
the classes are about.
I would be glad if someone volunteers to do this, but this needs a lot
of time for someone without any experience in using Qwt.
> What the maintainership load will be, only you can tell. I really
> if those modules would attract attention and contributions, generating
> reviewing load for you.
Qwt has been developed in the public since many years so I don't expect
that anything changes only because of being on qt.project.
The only new input I can imagine would be coming from the regression and
> I don't know whether the bug report rate would require
> the Qt Project QA to require the Maintainer to take action (the
> responsible for the module being "always ready for beta" and the
> release requires the last released version to work with the all the
> released versions).
I never had more than 10 open bugs in the Qwt bug tracker in the
complete history of the project.
I'm used to fix them in the next days after they have been reported.
> So this is really up to you and the question is: do you want to?
But not only me - the integration of an existing project like Qwt with
its size and a history of many years might be a challenge for qt-project
But Qwt is not only a library - it is also a user base to take care of.
I don't know myself how many there are, but sourceforge reports 1200
downloads - only for the last week.
There is one question left: the Qwt license is the LGPL with some
exceptions allowing static linkage without having to show the
These exceptions were ( are ? ) useful in combination with a commercial
I don't care much about these exceptions, but changing the license ( or
the copyright ) means to get permission from people that got lost many
More information about the Development