Seeking Clarification from Apple regarding UITextInput Selection handling

[Oliver Drobnik]

This is the 5th way that I am trying to get Apple to clarify how we developers are supposed to get UITextInput to add the standard gesture recognizers to your custom edit view. The other 4 ways are:

- mailed the developer evangelist with a detailled description of the disconnect between what we developers think is provided and what Apple thinks they are providing
- submitted a Radar
- submitted a tech incident
- mailed an engineer who at WWDC told me he is working on text input


If you implement UITextInput on your custom view and - say - use CoreText to render the text you get to a point where you can draw your own cursor and selection/marking and have that fully working with the hardware keyboard. If you switch to Japanese input then you see the marking, but there's something curious: if you long press into the marking you get the rectangular system loupe and selection handling without having to deal with the touches yourself.

What I don't get why we would have to implement our own touch handling for the selection, draw our own loupes etc. It's working for marking! So what do I have to do to get the standard gesture recognizers added to my custom view as well?

the one sample on the dev site only has a comment about that user selection would be outside the scope of the sample. Which would indicate that indeed you have to do it yourself.

I don't think that it is in Apple's interest that all developers doing their own Rich Text editor class keep doing their own selection handling code, let alone custom drawing of the round and rectangular loupes?! Granted you can try to reverse engineer it such that it comes really close, but that might give users a strange feeling if the selection mechanics differ ever so slightly.

I found that developers are split in two groups:

1) rapes UIWebView with extensive JavaScript code to make it into an editor
2) painstakingly implements the selection mechanics and loupe drawing themselves

So what is the solution here? Keep submitting Radars until Apple adds this missing piece? Or is this actually already existing (as claimed by the aforementioned engineer I met) and we are just too stupid to find how to make use of it, instead resorting to doing everything (but marked text) manually?

Even the smart guys at Omnifocus seem to think that the manual approach is the only one working. This makes me sad, you get such a great protocol, but if you implement it you find it severely crippled. Maybe even intentionally?

[harrytheshark]

I've found the same thing, it appears to be very much incomplete - half of the methods don't even get called.

Adding the loupe stuff is a pain, but it's much easier than it first appears to be - I've got it working quite nicely.