Re: RTL PUA?

From: Michael Everson <everson_at_evertype.com>
Date: Fri, 19 Aug 2011 16:29:30 +0100

On 19 Aug 2011, at 16:05, Shriramana Sharma wrote:
> On 08/19/2011 08:03 PM, Michael Everson wrote:
>> On 19 Aug 2011, at 15:13, Doug Ewell wrote:
>>
>>> The PUA is supposed to be a free and open sandbox, without reserved
>>> or allocated zones.
>>
>> Nevertheless, inherent directionality is something that computers
>> take notice of. There would be no harm in having a RTL PUA area.
>
> To set restrictions on the usage of the PUA by specifying specific LTR
> and RTL ranges would be to undermine the definition of the PUA, IMHO.

The PUA *already* defines its characters as LTR. That's been done. It is *part* of the definition and functionality of the PUA. It's irrelevant whether it "should be" or not. It *is*.

Another requirement is to be able to exchange private characters which function correctly in an RTL environment. The sensible way to achieve this is to add new a new PUA area which defines its characters as RTL.

> As Doug correctly pointed out, anybody can do anything with PUA codepoints -- render them LTR, RTL or even top-to-bottom or bottom-to-top or boustrophedon if they can.

Please write me a rendering engine that will correctly process http://www.evertype.com/standards/csur/engsvanyali.html on the Mac OS, Linux, and Windows. Thanks.

>>> My question would be why the PUA is designated as 'L' by default at
>>> all, instead of, say, 'ON'.
>
> As I already said, it would seem then that the correct (or at least, *best*) BC for the PUA is ON, at least implying that the Unicode Standard itself doesn't specify any directionality for these characters, whether that is the original intention of BC=ON or not (because it might be to actually assert that they *are* *neutral* [which means exactly what I'm not sure as they *have* to be written in *some* direction...]).

But the standard already *does* define their directionality, and that can't just be "undone" because you think it would be more logical. It is equally logical to say that it is perfectly reasonable for the current PUA to be LTR (which it is) and to add a new PUA which is RTL.

>>> So your private agreement, in addition to specifying the meaning of
>>> your PUA characters and probably some sample glyphs, can also
>>> specify their properties, overriding the default properties.
>>
>> Gods know I wouldn't have any idea how to get my operating system to
>> honour such a declaration.
>
> See my other mail for differences in abilites. You can do a font, others can't. Others can do rendering engines, you can't. C'est la vie!

Oh, come on. *Anybody* can download Fontforge or buy Fontographer and draw some glyphs. You could even do so, with no particular training. Your glyphs might suck, but they would render and your fonts would work. That is *completely* different from being able to write software that overrides the system-level expectations of directionality on the Mac OS, Linux, and Windows. Be realistic.

>>> There's a lot of misinformation and FUD about the PUA, and
>>> unfortunately I expect at least one response of the form "The PUA
>>> is evil, don't use it," which accomplishes very little.
>>
>> I just think we need some PUA that's RTL.
>
> Please don't take this amiss, but I would like to hear more than just a restatement of your opinion (which is clear enough), more especially in the nature of sufficient arguments as to how that would be meaningful or helpful to allocate RTL PUA, or as to how it would *not* undermine the very *definition* of the PUA! :)

The very *definition* of the PUA specifies that all those characters *are* LTR. After two decades, you want to undo that? I don't think so.

Michael Everson * http://www.evertype.com/
Received on Fri Aug 19 2011 - 10:31:47 CDT

This archive was generated by hypermail 2.2.0 : Fri Aug 19 2011 - 10:31:53 CDT