Re: Bug in NSAttributedString (?)

[Richard Frith-Macdonald]

On Sunday, November 3, 2002, at 04:40  pm, Pierre-Yves Rivaille wrote:

> The implementation of the shouldChangeTextInRange:replacementString: 
> method clearly follows the apple doc. I believe the problem is in the 
> replaceCharactersInRange:withString: method. This method should not 
> call shouldChangeTextInRange:replacementString:. The apple doc says 
> "This method must be invoked at the start of any sequence of 
> user-initiated editing changes" which *probably* means that it should 
> *only* be called on user-initiated changes.
> The same problem occurs in GNUMail. When no message is selected, the 
> textview is sent a setString: @"" message, and this message is ignored 
> because shouldChangeTextInRange:replacementString: responds NO.
>
> I believe the same problems happens with -setFont:ofRange:, 
> -setColor:ofRange: -setAlignment:ofRange:.

Certainly I agree there is nothing wrong with 
shouldChangeTextInRange:replacementString: and we should not change 
that just because if appears to fix the immediate problem.
The idea that shouldChangeTextInRange:replacementString: should only be 
called for user initiated changes sounds plausible, but raises the 
question of what should be considered 'user initiated'.  Perhaps any 
fairly direct response to
a mouse operation or a paste?