From: verdy_p (verdy_p@wanadoo.fr)
Date: Wed Nov 05 2008 - 22:47:16 CST
> Message du 06/11/08 00:02
> De : "philip chastney" <philip_chastney@yahoo.com>
> A : "Unicode Mailing List" <unicode@unicode.org>, "Doug Ewell" <doug@ewellic.org>
> Copie à :
> Objet : Re: Boustrophedon (was: Re: Question about the directionality of "Old Hungarian" (document N3531))
>
> --- On Tue, 4/11/08, Doug Ewell <doug@ewellic.org> wrote:
> From: Doug Ewell <doug@ewellic.org>
> Subject: Boustrophedon (was: Re: Question about the directionality of "Old Hungarian" (document N3531))
> To: "Unicode Mailing List" <unicode@unicode.org>
> Date: Tuesday, 4 November, 2008, 2:07 PM
>
> Q: Why is this thread like boustrophedon itself?
> A: Because it goes in two different directions.
>
> It would be really neato if we could split the "principles of
> boustrophedon" discussion off into a separate thread, called, oh, I
> don't know, maybe "Re: Boustrophedon", and leave the present
> Subject line for discussions that really do have to do with
> Old/Szekler
> Hungarian.
>
> fair enough --- may I start with a query?
>
> p218 of 'Reading the Past' (British Museum Press, 1990) says, "Some
> early Greek and South Arabian texts were written boustrophedon [...].
> In such inscriptions the letters are often reversed to face the
> direction of writing."
> that's "often", not "always"
> unfortunately, the illustrations provided all show mirrored lettering
> so, are there any examples of boustrophedon writing where the letters are NOT mirrored?
I've not seen such examples. But if implicit mirroring is the rule to apply to Boustrophedon presentation style,
then it shuold apply be default to all scripts (including Latin and Arabic or Hebrew), meaning that situations with
double-mirroring will occur. Such implicit mirroring does not have to be handled by the BiDi engine but by a higher
layer that swaps the direction of text and mirrors the direction computed by the BiDi algorithm.
In this case what will happen to the situation in a paragraph like "Latin-text-1 Boustrophedon-text Latin-text-2"
when there's a line break in the middle of the Boustrophedon-text?
- For the Bidi algorithm, Boustrophedon just behaves like normal LTR text, so BiDi does not produces any mirroring
or change of direction any where in the paragraph. But then the Boustrophedon text is mirrored on the second line
as well as the effective baseline direction (that will start on the left).
- Then comes the second Latin text: the baseline goes in the wrong direction for Latin, which has a strong LTR
direction so the Latin letters have to be reversed on this second line, and must be double-mirrored (i.e. not
mirred at all). If such effect is not desired, the only safe way is to present the Latin lext in RTL style with all
letters mirroed by the preceding Boustrophedon context.
- One way to force the double-mirroring (and restore the strong LTR presentation seems to be that of using a BiDi
LTR override control between the end of Boustrophedon-text and the second Latin text. This will not change the
result computed by the BiDi algorithm, but it will hint the Boustrophedon-capable renderer (that will use the
result of the BiDi algorithm) to NOT maintain the mirroring and reversal of Latn letters after it. For this to
work, the BiDi algorithm MUST be capable not only to indicate the status of letters (mirrored, reordered) but ALSO
if such information is infered or enforced due to a strong character property (or BiDi override control): with such
control present, the diretionality is enforced by the BiDi algorithm and the Boustrophedon upper layer MUST not
reorder the letters and MUST NOT mirror them again, even if the current baseline direction (including the position
of the start margin) is wrong; instead it will render the span of text after this control on the left of the
Boustrophedon text, without isnerting a line break, and this span will be right-aligned to the end of the RTL span
of Boustrophedon on the same line. If a line break occurs in this Latin text after the Boustrophedon, the normal
left margin will be used for the rest of the Latin text, as well as the normal LTR direction of the baseline, and
no mirroring.
In all these cases, the BiDi algorithm does not need to be changed, it must just be applied as a layer below the
Boustrophedon presentation upper layer. And scripts normally written in Boustrophedon style do not have to be
handled differently by the BiDi algorithm as normal LTR scripts (Latin, Greek, Cyrillic, Brahmic abugidas,
ideographic, ...)
This archive was generated by hypermail 2.1.5 : Wed Nov 05 2008 - 22:49:31 CST