Re: [lmi] Custom wxArtProvider affects html help toolbar buttons

[Greg Chicares]

On 2009-03-18 07:33Z, Vaclav Slavik wrote:
> On Tue, 2009-03-17 at 18:55 +0000, Greg Chicares wrote:
>> and so my question is whether the 'wxDefaultSize != size' case matters.
> 
> In practice, it almost certainly doesn't -- nobody is likely to call it
> in such way in the current version of wxWidgets code.
> 
> In theory, it does matter: requested size is a part of
> wxArtProvider::CreateBitmap() API and any implementation should respect
> it. OTOH, there's fallback code in wxArtProvider that ensures that if
> CreateBitmap() returns bitmap of incorrect size, it is rescaled to the
> requested one (at some lost of quality).
> 
> In other words, requested size should be respected (irrespective of wx
> version), but the only penalty for not doing so is a small loss of
> quality in some very rare situations that probably don't occur in LMI in
> the first place.
> 
>> (Where would a nondefault size come from, if not from GetSizeHint()?)
> 
>>From a wxArtProvider caller that does something unusual and needs a
> bitmap of specific size. The only real-life examples of such situation
> that I can think of right now are (a) $(wx)/samples/artprov and (b)
> Poedit's attention bar (similar to
> http://webct.ipfw.edu/troubleshooting/images/ie-information-bar.gif, but
> with standard "warning" icon - i.e. downscaled version of
> wxART_WARNING). Neither of these is relevant to LMI, of course.

Fascinating. Thanks.

We'll keep the slight extra complexity in lmi. The runtime cost is too
small for anyone to notice. Robustness is good. I can well imagine that
the answer to someone else's arcane question, somewhere else, would be
that lmi provides the only known example of some other exotic wx usage.