On 3/29/2014 9:18 AM, Mojca Miklavec wrote:
On Sat, Mar 29, 2014 at 7:51 AM, Jan Tosovsky wrote:
On 2014-03-28 Hans Hagen wrote:
On 3/28/2014 7:23 PM, Jan Tosovsky wrote:
Btw, has anybody any idea where the following 'dotlessi' patch has
been lost?
http://www.ntg.nl/pipermail/ntg-context/2013/076306.html
It was working in December in ConTeXt 2013.12.20 / LuaTeX 0.77.0 on
Windows 7
because it was not okay (too much fuzziness and dead slow too) ...
Sad to hear that. I thought the patch was about removing duplicates when
building the initial glyph table and didn't expected these severe side
effects...
I don't know if we are talking about the same "issues" (there have
been plenty of problems with building the font database), but building
the font database dropped from 45 minutes to less than a minute on my
machine. (OK, half an hour was due to a single font.) Reading the
whole font for each font in the system when building the database was
an overkill.
Some problems could be fixed per-font during font loading though. I'm
just speculating because I don't know the code. But doing an O(n^2)
comparison of glyph names on a font that covers the whole Unicode
should be avoided in principle.
another aspect is that once a glyph would be tagged as 'foo' and that
goes though the system it gets harder to figure out what it originally
was; using indices throughout the system is more robust (accompanied by
some lua magic) but that is a rather intrusive solution (i have some
ideas on that and will experiment with that later this year as it is not
something we want to mess with around tl freeze time)
Hans
-----------------------------------------------------------------
Hans Hagen | PRAGMA ADE
Ridderstraat 27 | 8061 GH Hasselt | The Netherlands
tel: 038 477 53 69 | voip: 087 875 68 74 | www.pragma-ade.com
| www.pragma-pod.nl
-----------------------------------------------------------------
