Re: [TCLCORE] Tip #248 (tile)

[<geo...@xm...>]

Quoted Vince Darley <vi...@sa...>:

> On Thu, 9 Jun 2005, Donal K. Fellows wrote:
> [Lots of very sensible comments snipped]
>> Vince:
>> > I expect that some of those deficiencies ought to be rectified before Tile
>> > should be distributed with Tk.  In particular I recall some very
>> > non-Tclish behaviour of the inability to redefine a style without exiting
>> > Tk.  I expect there are many more such issues -- areas where Tile is
>> > slightly backwards incompatible without a good reason.

> Given this position, I hope you can understand why I'm somewhat more
> cautious about just giving Tile the stamp of approval as-is, without
> really understanding what its limitations/incompatibilities are, and which
> of those are considered 'bugs' and which are not.

Not having the ability to redefine a style is problematic.  For instance a
person could someday create a Tile-style creation program for people that
aren't programmers.  This could be a case of a situation that is impossible
to handle without starting a new wish for each minor style change.  It would
hurt interactivity in that case.  I understand that Tile is complex, and some
of the native APIs it uses are too, but it would be nice to allow for dynamic
style redefinitions to enable such programs.  Some of us also use editors
that save and update running processes (via send or comm), so it could be
useful in those situations too.

GPS