As it was pointed out in comment #43 for bug 25526 (https://bugs.webkit.org/show_bug.cgi?id=25526#c43), and properly tracked down through bug 50536, GTK is about to stop being the only platform implementing caret browsing mode (well, Mac has the VO cursor, which is a similar thing but a different implementation) since Qt is going to support it too. Hence, it would be interesting to have all those Layout tests out of platform/gtk, where they normally are nowadays.
Current proposal to fix this bug would therefore be to do the following:
* Move all the caret-browsing related Layout tests out from platform/gtk to editing/caret-browsing. (I think this would be a good place to put them all, but if somebody has a better suggestion it would be awesome to share it)
* Put the moved tests in the Skipped files for all platforms but GTK, which would be the only one initially passing all of them.
* Further tests related to caret-browsing should be placed under that path, and initially skipped in all platforms but the one adding them.
As there's currently a couple of patches pending on review which would add these kind of tests under platform/gtk (see bug 27048 and bug 25533), my personal proposal it's to keep the current policy of adding them as platform specific, and after both of them get integrated, providing a patch to fix this bug but moving all the tests together to the right place. That's why I'm settting this bug as dependant on those.
This way we have the double benefit of not messing things up in patches currently on review while, at the same time, leaving room to think about this bug and the proposal itself (such as if the right destination path should be editing/caret-browsing, or somewhere else).
Depending on bug 50536 as well, as the patch for it also involves moving a test from platform/gtk to editing/selection. Hence, it would be nice we wouldn't be proposing a patch for this one until that one gets integrated too, to avoid conflicts.
Btw, I wonder whether that could mean that perhaps it's better to move the patches to editing/selection instead of editing/caret-browsing, as proposed in this bug's description. IMHO, I think it's better to create a editing/caret-browsing directory to differentiate tests that are just checking the caret position in certain situations (those usually under editing/selection/*caret*html) from those testing caret navigation in caret browsing mode, which is what this bug is actually about.
(In reply to comment #0)
> As there's currently a couple of patches pending on review which would add these kind of tests under platform/gtk (see bug 27048 and bug 25533), [...]
My mistake, sorry. bug 27048 shouldn't be blocking this one since it provides an GTK-specific accessibility layout test (even if it's related with caretbrowsing).
Removing from the list of dependencies. Sorry for the hassle.
I need to get more quality sleep time... it was bug 25533 what this one should be blocking, not 25333 :/
I like editing/caret-browsing