Re: Patch: Bug 50278

[Mike Kestner <mkestner ameritech net>]

Owen Taylor wrote:

> And since you'll almost certainly need to adjust the value in
> more code, there is little advantage to putting the value in
> the constructor.

I think providing a hardcoded page=10*step is consistent with the goal
of this helper API, but value cannot be defaulted meaningfully.  Some
spins start at min.  Some start at 50%. In my random sampling analysis
of lxr, I don't recall any situations where the initial value provided
in the gtk_adjustment_new call was altered outside of callback code.  If
we don't provide the value parameter, we are back to a two call
construction paradigm, which isn't significantly more compact than the
existing mechanism.
 
> As Tim said, We don't include the spin_button != NULL checks any more
> since IS_SPIN_BUTTON does that.

In my node, I've removed a couple dozen identical legacy calls as well. 
:-)
 
> This is pretty ugly even for testgtk... why don't you add a quick
> enum here.

Heh, of course I yanked all that from existing spin code in testgtk. 
:-)  I'll try to clean the whole spin section up a bit.
 
> Why don't you send another version of the patch 

I'll hold off for a day or two to give people a chance to consider your
comments.  Some consensus regarding the value parameter would be a good
thing too.

Mike