Re: problem with r2l's state attributes

[Tzafrir Cohen]

[ Sorry. Been away from the list for a couple of days]

On Fri, 3 Aug 2001, mulix wrote:

> On Fri, 3 Aug 2001, guy keren wrote:
>
> > another problem, with the way we use the file to store the states for r2l
> > (enabled/disabled and r2l base mode) is:
> >
> > 1. when we disable r2l, we erase the file, and thus 'forget' the biditext
> >    base directionality. thus when later we enable r2l, it's enabled in the
> >    default mode (neutral) - not a good idea.
>
> tzafrir? should the biditext base directionality be persistent accross
> r2l changes?

I figure that in the cases where a direction is forced, then if you
temporarily disable biditext, you would expect it to have the same forced
base direction. This seems like a sane behaviour

>
> > 2. specific to r2llib - when one sets the biditext base mode, you
> >    automatically enable r2l (implicitly, by creating the file). this isn't
> >    a great idea either.
>
> why not? is there a meaning to a base biditext direction when we are in
> l2r mode? doesn't setting the base biditext direction imply that we want
> biditext activated?
>
> > perhaps we need to use 2 files - one for enable/disable, and one for
> > directionality? e.g. rev.state and rev.bt_state ?
>
> possible, and very easy to implement. personally, i dont care either
> way.

Actually, of switches come cheaply, then maybe it is better to have three:

disable/enable

       neutral/set base dir

              rtl/ltr

Anything with more than two states seems to me as less intuitive.

>
> > sometimes i just wish to state that _when_ r2l be enabled - it should be
> > enabled in rtl mode, but yet i don't want it to be enabled right now.
>
> i'm not sure this is a problem the library should solve, but rather the
> application. we should probably follow the principle of least surprise
> here - do you (guy) as the application author find the current behaviour
> non intuitive?
>
> > 3. another problem with r2llib's interface - either you always use r2l and
> >    l2r, or you always use rtl and ltr - don't mix the 2 - i keep getting
> >    confused! :)
>
> function names begin with r2l_xxx_xxx(). the r2l opaque token each
> function receives is named rtl, since i dislike variables with digits
> int their names.

I've already done to many typos with this '2'.

Anybody else for a mass renaming s/r2l/rtl/ ?

-- 
Tzafrir Cohen
mailto:[EMAIL PROTECTED]
http://www.technion.ac.il/~tzafrir