Re: padding cleanup

[Havoc Pennington <hp pobox com>]

Hi,

On Wed, Sep 1, 2010 at 2:03 PM, Shaun McCance <shaunm gnome org> wrote:
> Well, all of the packing functions use start and end, but I
> guess that's just to make the term orientation-neutral.
>
> Looking through the docs, I do see properties like left-attach,
> left-margin, and left-padding. So it doesn't make sense to use
> start and end unless we switch all terminology. That's probably
> just pointless API churn.
>
> Side note: browsers are starting to provide properties using
> start and end:
> http://help.dottoro.com/lcedmdkb.php
>

Well, what does GtkAlignment do already with its padding?

I think the question is, does GTK just flip left and right, or does
GTK let you choose to force left/right regardless of text direction
and then also let you "virtualize" with start/end. I'd guess the
browsers are not comfortable / unable to just flip all lefts and
rights so they are adding new props for auto-flipping padding.

The problem is if you have something called left or right which does
NOT auto-flip, people will use it, and then be broken in RTL.

I'd almost be inclined to have foo-left, foo-right which flip and then
_if there's a use-case_ foo-left-in-all-text-directions and
foo-right-in-all-text-directions which don't flip. But I don't know
much about it, honestly. However it works, the obvious, default thing
app developers will do probably ought to work by default in RTL. I
kind of suspect padding-left and padding-right are the obvious default
thing app devs will use.

Havoc