viewRequestID regression and mixin issue

[ahocevar]

r10755 brings a regression, because the code shuffling needed for making Tile.Image.IFrame a mixin caused an unintended change in execution flow: renderTile needs to update the viewRequestID every time, and not just when a new imgDiv is created.

The way Tile.Image.IFrame is mixed in will cause the mixin to always be applied, unless the maxGetUrlLength is explicitly set.

The attached patch fixes these regressions. All tests pass. Thanks for any review

[ahocevar]

Altough the patch may look scary at first sight, it only does two things:

• in Tile.Image, it applies the mixin after calling the super's initialize method, so we can extend "this" rather than "options".
• in both Tile.Image and Tile.Image.IFrame's initImgDiv methods, the viewRequestID will be updated regardless of imgDiv being null or not (i.e. we don't return at the top of the method any more when imgDiv is not null).

[tschaub]

Since you've got your head deep into this (and we're far from a release), I'll trust this does what you've said. Would it be possible to get a regression test at least on the viewRequestID bit? In either case, thanks for making this work.

[ahocevar]

r10773, which includes a regression test for the viewRequestID bit