Krazy Check for Missing toolTips and whatsThis

[Allen Winter]

On Sunday 07 September 2008 11:00:22 Parker Coates wrote:
> On Tue, Sep 2, 2008 at 16:43, Allen Winter wrote:
> > I'm looking for feedback on how well this checker works.
> > It should process C++, Qt designer (.ui) and KConfigXT (.kcfg) files.
> 
> So I added the test last night and have the following comments.
> 
Nice, thanks for the feedback.

> - The checker currently complains about QLabels not having tooltips or
> whatsthis text. I'm not sure where the usability folks stands on this,
> but it was my understanding that these extra bits of help information
> should be added to control widgets, but not necessarily to the labels
> describing them. Others may disagree, but I would recommend disabling
> the check for QLabels.
> 
I don't know if QLabels should be checked either.
Any usability folks out there wish to comment?  Should QLabels have tooltips/whatthis?

> - The checker complains that widgets controlled by KConfigXT (named
> "kcfg_*") are lacking tooltips and whatsthis text. Since these widgets
> pull the tooltips and whatsthis text from the .kfcg, they're not set
> explicitly in the C++ source. I would recommend disabling the check
> for variables starting with "kcfg_".
> 
I can do that, sure.

> - My application has two .kcfg files, one is used in the traditional
> sense to store application settings and to hook up to a settings
> dialog. The other is used to define a custom filetype used by my game
> to load store and load different rulesets. This second .kcfg file
> contains no tooltip or whatsthis tags because it has no association
> with any GUI. Is there a way to tell Krazy not to run a particular
> check on a particular file? Or vice versa, to run a particular check
> only on a particular file? I realise I could make this work by
> splitting the files into separate directories and adding .krazy files
> to each, but that seems a bit kludgy.
> 
Yep, anywhere in the .kcfg file put the comment line
<!--  krazy:excludeall=tipsandthis -->

> Anyway, I like the checker and I think it makes sense that one has to
> opt-in. I hope the above is useful.
> 
> Parker
> 
> 
> P.S. Sorry that this email breaks threading, but I subscribed to the
> list after the original email was sent, so I had to fake the reply.
> 
>