On Wed, Sep 19, 2018 at 4:53 PM, Huw Davies huw@codeweavers.com wrote:
I think implementing LBS_NODATA should come before adding LB_SETCOUNT code that will need to be changed again. Also, adding a test to show that LB_SETCOUNT fails if the style is LBS_OWNERDRAWFIXED, for example, wouldn't hurt.
Huw.
To be honest, we should just fail SetCount if LBS_NODATA is not set, but maybe later not now, if tests prove this is Windows' behavior, which I'll check in a bit. Even if Wine doesn't support it, it still "works" currently (but very slow and suboptimal, there's some bugs around about it), so I think it should be left with this wrong behavior for now.
BTW I think you meant OWNERDRAWVARIABLE right? LBS_OWNERDRAWFIXED should work (and is actually required) for LBS_NODATA (and thus, LB_SETCOUNT), or I'm missing something.
I guess for now this patchset should be rejected until LBS_NODATA is implemented which would require a totally different patchset anyway, or I find some other behavior with tests.