[thanks for sending this to a public list, Keith]
On Thu, Oct 19, 2000 at 03:07:30PM -0700, Keith Packard wrote:
>
> > > > The spec says you get "an error" when the following are true:
> > > >
> > > > The widget is managed.
> > > > The widget's parent is realized.
> > > > The parent's class is not a subclass of compositeWidgetClass.
>
> > I'm getting tons and tons of bug reports from Debian users of my beta
> > XFree86 packages over this issue; if I release official packages like this
> > that will turn into a flood. Please, please work something out with Keith.
>
> I'm pretty stuck at this point; the spec is rather clear on this issue and
> the implementation has always been broken. X usually focuses on the spec,
> but in this case:
>
> 1) the implementation has always been broken.
> 2) the implementation was written *before* the spec.
> 3) the implementation is well behaved (and useful).
> 4) changing the implementation breaks most apps.
>
> Given the haphazard development of the spec, I'm tempted to ignore it in
> the case of Xt and focus on making applications function again. I'd also
> suggest editing the spec and making the existing behaviour standard.
>
> This will cause errors when VSW5 is run against our Xt version.
>
> I don't expect to hear any complaints on this list for breaking compliance
> with the Xt spec; unless people have serious objections, I think we should
> revert Xt to the original code.
I have no objections to reverting the change, or attempting to get the spec
modified. Thanks for suggesting a resolution I think we can all live with.
Ordinarily I don't like it when implementations break a spec, but I think
this may be an exception.
--
G. Branden Robinson |
Debian GNU/Linux | If encryption is outlawed, only outlaws
[EMAIL PROTECTED] | will @goH7OjBd7*dnfk=<q4fDj]Kz?.
http://www.debian.org/~branden/ |
PGP signature
