From: Julian Bradfield (jcb+unicode@inf.ed.ac.uk)
Date: Wed Aug 19 2009 - 04:47:59 CDT
David Starner wrote:
>This decision, either way, is completely backwards compatible.
>Provided you don't update to the latest fashion, your system will work
>exactly the same way it always has. The only question is whether the
>latest fashion, which you here deny an interest in, will be based on a
>concern for running on your system.
Yes, but it won't work correctly, just as it doesn't now. The original
unification is (IMO) a bug. Disunifying is a bug fix, not a new feature.
>[if computers are] really a once a decade luxury, I seriously doubt it matters to
>you enough that you care about the fine points of IPA typography.
OK, let's forget the poor. There still remain the rich and scientific.
>> For the last (almost) twenty years, I've been using the same core
>> application. In the last ten years, that application has changed
>> hardly at all. It predates Unicode, but now sort of supports it; it
>> far predates Pango, and I see little prospect of anybody doing the
>> massive job of porting it to Pango in the foreseeable future - quite
>> apart from the fact that that would introduce lots of dependencies on
>> rapidly changing software.
>
>How poorly designed can you get? Even twenty years ago, they knew
>better than to intertwine the interface with the core code. If it were
>properly coded, it wouldn't be a massive job.
Er, it's an editor. The interface is part of the core code.
Also, twenty years ago, they knew that in some cases, if you tried to
separate things out in the approved fashion, the performance would be
unacceptable. Do you remember when Microsoft tried to Do the Right
Thing and move the GUI stuff out of the Windows kernel? They had to
move most of it back in again to get an acceptable user response.
The reason that XEmacs's display code is the code from hell is that it
had to run with acceptable latencies on machines a hundred times
slower than today's - I first used it on a 68040 - while still
providing display functionality as rich as, or richer than, Pango/GTK
do now.
The display module is in fact well separated from the rest of the
editing code - but it's an extremely complex module, since it does
the same job as the GTK text display widgets and all their underlying
code.
>I'm telling you that if you want the fanciest new features, you need
>to be prepared to update your code for them. If you're happy with the
I don't want a new feature. I want a bug fixed.
>Unicode, especially when that feature is largely cosmetic.
If you've followed this discussion at all, you should appreciate that
the argument is not cosmetic. Ignoring Michael's fetish for setting
books using only one font, typography is not an argument for plain
text. The argument is that IPA and Greek letters are *logically
separate* letters, and should therefore be encoded separately, for the
sake of data processing on them. Display is only one form of data
processing. Variation selectors are, as Michael says, a
"pseudo-encoding" which allows one to separate the letters for data
processing purposes in a way that preserves readable plain text
display (provided your systems at least know to ignore VS) - but it's
a hack, and doesn't even work all that well, because most
Unicode-based algorithms will treat a variant as equivalent to the
base character.
-- The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336.
This archive was generated by hypermail 2.1.5 : Wed Aug 19 2009 - 04:52:16 CDT