Re: textview iter_inside_word bug?

[Havoc Pennington <hp redhat com>]

Evan Martin <martine cs washington edu> writes:
> Inserting large blocks of spaces somehow confuses
> gtk_text_iter_inside_word(), having it report you're within a word even when
> you haven't even typed any letters.
> 
> To reproduce:
> run attached test case, click on the text box, hit space about ten times
> and watch the console.  This prints the status of the iterator at the
> insert position every time you type.
> To understand the output, read the code.  It's pretty simple.

Can you put the test case in bugzilla so it doesn't get forgotten?
 
> (This may be related, too:  the insert-text signal on a GtkTextBuffer is
> incorrectly documented as taking a GtkTextIter as an argument, when in
> fact it appears to take a GtkTextIter*.  If you try to use a passed
> GtkTextIter you get a pile of errors, while if you use a GtkTextIter*
> everything works as expected with the exception of this bug.

The signal docs should be autogenerated, so this is a gtk-doc bug
presumably.

> Additionally, a cursory glance at the code shows the insert_at_cursor
> signal on a GtkTextView [View, not Buffer] doesn't seem to be used for
> much.)

insert_at_cursor is a keybinding signal. It's used to implement
M-space for the Emacs key theme, for example.

Havoc