Hopwood wrote:
> There is only one use of CGJ: to link characters into a single grapheme
> cluster
Yeah, maybe, but I'm still lingering in the "shouldn't have been encoded"
camp. It happens to me every time we encode another control-like
thingamabob to solve a dubious "plain text" problem...
I would not want to see CGJ become an all-singing all-dancing monster of
implementation compexity masquerading as a panacea for plain-text markup.
But I'm probably too late to stop the onslaught anyway. People are gonna
continue to dream up "one more little rule" to tack onto the list of little
rules, and before you know it, we'll end up using it for, to paraphrase
Hopwood, extending diacritics across an arbitrary number of camels'
neuroses.
It's not appropriate to go down that road in plain text. Every tweak of
interpretation and every newly allocated control-like character is another
step toward fiery abyss, away from the true purpose of the character
encoding; another incursion across the tracks into the shady side-streets
and detritus-ridden alleyways of Markupville.
> less complicated than adding all the double diacritics that would
> be needed [...]
Such as what? In the 12 years that I've been hanging around here, nobody
has come up with more than a handful that really _are_ used in what might
plausibly be called "plain text", and mostly for obsolete dictionaries. How
many more of these beasts are lurking in the shadows waiting to pounce?
Probably not many.
In any case, I still maintain that this CGJ -- along with other similar
plain textish markup-like mechanisms -- is going to lead people down the
garden path into a hideous trap of ever increasing plain text complexity.
The current thread is an example of the growing problem. Got a stumper?
Just throw another control code on the fire and the problem's solved.
By the way, has anyone ever actually implemented this CGJ in proto-type
and run it through some user testing?
Rick
