`face' widget type

`face' widget type

[Richard Stallman]

I got rid of the confusing double-levels of hiding,
and the double magic button, by redefining `face'
in terms of `editable-field'.  I also put in a "sample"
to show what the chosen face looks like.

It works except for one thing: when I set a different face name,
the sample does not update.  How can I make it update?

Re: `face' widget type

[Per Abrahamsen]

Richard Stallman <rms@gnu.org> writes:

> It works except for one thing: when I set a different face name,
> the sample does not update.  How can I make it update?

I don't have time to find out.

Re: `face' widget type

[Richard Stallman]

    I don't have time to find out.

Will you have time in the future?
Can you tell us some advice?
I think nobody but you really understands this code.]

Re: `face' widget type

[Per Abrahamsen]

Richard Stallman <rms@gnu.org> writes:

>     I don't have time to find out.
>
> Will you have time in the future?

Things I don't do right away tend not to be done at all, or many years
later.

> Can you tell us some advice?
> I think nobody but you really understands this code.]

My advice would be to delay any improvement (except for fixing
regressions) until you are at the *start* of a development phase.  The
code *is* complex, and there is some non-trivial dependencies, so any
changes should be done at a phase where there is plenty of time left
to find any bugs they might uncover.

I.e. now is entirely the wrong time to start improving the customize
UI.

Re: `face' widget type

[Stefan Monnier]

> I.e. now is entirely the wrong time to start improving the customize UI.

Couldn't agree more,


        Stefan

Re: `face' widget type

[Richard Stallman]

    My advice would be to delay any improvement (except for fixing
    regressions) until you are at the *start* of a development phase.

We're talking about three kinds of cases: fixing old bugs, fixing new
bugs, and changes that are not bug fixes.  You're grouping the first
case with the third; I think it belongs with the second.

I think it is misleading to describe bug fixes as "improvements".
That word suggests something optional, and bug fixes are not optional.

Re: `face' widget type

[David Kastrup]

Richard Stallman <rms@gnu.org> writes:

>     My advice would be to delay any improvement (except for fixing
>     regressions) until you are at the *start* of a development phase.
>
> We're talking about three kinds of cases: fixing old bugs, fixing new
> bugs, and changes that are not bug fixes.  You're grouping the first
> case with the third; I think it belongs with the second.
>
> I think it is misleading to describe bug fixes as "improvements".
> That word suggests something optional, and bug fixes are not optional.

There are different severities of bugs, though.  Fixing a user
interface design inconsistency that leads to bad results in marginal
cases is to be weighed against the likelihood of the respective fix
introducing more severe problems that might go unnoticed before the
release.

A redesign might in cases be necessary to actually fix a minor bug
properly.  It might make sense to move that kind of fix to after the
next release, and instead document the problem for the current
release.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: `face' widget type

[Richard Stallman]

      Fixing a user
    interface design inconsistency that leads to bad results in marginal
    cases is to be weighed against the likelihood of the respective fix
    introducing more severe problems that might go unnoticed before the
    release.

This fix is localized.  It has a minor problem, which I asked Per to
help fix, but even with that minor problem it is nowhere near as
confusing as the old interface.