Le Sat, Mar 02, 2002, à 10:22:32AM -0600, Lars Clausen a écrit:
> I notice the appearence of a StringPrerenderer thingy. What's it do?
Been here for one year (at least). The goal is to make all (UTF-8) strings pass
at least once through it, so that any special (non-ASCII) characters could
be noticed and acted upon.
Currently (that is, before you last commit, which I haven't looked at yet),
this is used to build on the fly Postscript encoding tables (and used also
to be used by my Display Postscript experiment for the same thing). It also
sh/could be used to notice which glyphs we're going to need and put them
into some kind of prolog. I also use that to lazily encode the fonts,
instead of bulk-encoding everything (this is normally the default now in
0.88.1 and leads to faster EPS/PS). [*]
Of course, from what I understand of your recent postings, you've sort-of
re-implemented this in a second, parallel facility.
Obviously some merging will be necessary.
[*] building encoding tables is only half the job. Akira TAGOH rightly so
pointed out that we also need to locate the glyphs the printer is not likely
to know about, and put them in the prolog. So, your last commit, to the
extent it is able to download CJK/East-European glyphs, is a big step in the
right direction. I would have appreciated a lot if some public discussion of
this had been made before doing duplicate work (I deserve 80% of the blame,
for not having brought back to the list the discussion I had recently with
Akira on that topic).