PyQT Licensing and plugins/scripting

[Phil Thompson]

On Saturday 04 December 2004 10:30 am, Michael Sparks wrote:
> On Fri, 3 Dec 2004, Phil Thompson wrote:
> ...
>
> > The key is access to the Qt API. If your applications gives the users
> > access to the API then those users are developers and need their own
> > licenses. On the other hand if the API is sufficiently removed from the
> > Qt API then you shouldn't have a problem. Your API should restrict itself
> > to extending the capabilities of your application - the more general
> > purpose you make it, the more you risk a visit from the lawyers.
>
> What about, for example, a XUL processor? Suppose I wanted to rewrite
> Mozilla's front end to us Qt, I'd clearly need to implement a XUL
> processor. Obviously such a thing is possible to do with PyQT as well.
> Would I be able to do such a thing with a standard windows license for Qt
> and PyQt? Everything I've read suggests that this would not be possible.
>
> Users wouldn't have direct access to the Qt API, but they may have access
> to the aspects Qt system, assuming a XUL type system, including the
> ability to create new applications with new user interfaces (as one can
> using Mozilla, XUL and Javascript).
>
> ie one could envisage writing a wrapper around every part of the Qt API,
> and then expose that as an API - is that breaking the rules? I'd assume
> yes. Suppose then I simply change this to an XML processor (say a
> tokenising on)that when it gets a directive it simply calls the Qt API,
> and allow a user to change things in a config file. Is that too far? To me
> they seem equivalent.
>
> I don't tend to use windows much, if ever, and wouldn't want to do this at
> present. However, it's fairly close to something I would like to do under
> Linux (where this isn't a problem obviously), fairly close to the wind
> having read the commercial licenses I could see and it just concerns me
> that if I ever wanted to port such a system to windows I could get
> extremely stung (Suppose I was redistributing an executeable).
>
> It's a hypothetical question at present, due to using Linux, but it's
> (realistically) possible at some point it may become less hypothetical.

At this point I would take the coward's way out and say "ask Trolltech". As a 
matter of course I follow their model for PyQt (ie. if they are happy/unhappy 
then I'm happy/unhappy).

I suspect that you are right. Borland had to come to a special arrangement for 
Kylix - it wasn't just a case of buying a Qt license for each of their 
developers.

Phil