[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: [g-i] freefont vs. dejavu



On Thu, Mar 16, 2006 at 01:23:47AM +0100, Frans Pop wrote:
> On Thursday 16 March 2006 01:00, Davide Viti wrote:
> > This is (was) the idea (see [1]):
> 
> Looks good to me. Could you add the glyph ranges needed in the different 
> fonts (except freefont)?

I'm collecting suche ranges; I'll post them as soon as I make up my mind a little
bit more.
I'm considering to strip font files specifying the ranges to include rather than
the ranges that we want to exclude, what do you think about this?

 
> Please also consider if some common codepoints, like for numbers, general 
> punctuation and maybe accents should be included from the different fonts 
> in addition to their base ranges. This could help to keep visual 
> consistency within translations (only works if we implement changing the 
> default font on selection of a different language of course).
> 
> In some cases, like for CJK, we may want to keep latin characters too. For 
> CJK this has been discussed before (and AFAIK is already implemented in 
> the udeb).

total agreement on both points.
 
> > I wonder if we should consider switching to freefont whenever possible.
> > This would bring alot of advantages (both in size and simplicity).
> 
> That's basically what we've been doing so far, isn't it? 

yes. we switched to Dejavu for displaying Cyrillic and just found out 10
Cyrillic glyphs needed by the po files are missing :(

> Still, I think 
> that if a different font really looks better to native speakers than 
> freefont, we should in principle [1] use the other font [2]. 


> 
> [1] An exception could be if that font is much larger than alternative
>     fonts.
> [2] Within reason of course. We can't let a translator for, say, Dutch
>     decide he wants a different font. It's only acceptable for scripts
>     that have separate ranges in the UTF table.

Just to have an idea here's the situation. CJK have not been considered since
it's a separate case and no stripping is needed (I know Kenshi is taking
care of those)

DejaVuSans-Bold.ttf: 918 glyphs   [125188 bytes]
DejaVuSans-Oblique.ttf: 918 glyphs   [126872 bytes]
DejaVuSans.ttf: 918 glyphs   [141668 bytes]
nazlib.ttf: 303 glyphs   [68448 bytes]
nazli.ttf: 303 glyphs   [55941 bytes]
FreeSans.ttf: 2220 glyphs   [292904 bytes]
FreeSansBold.ttf: 1572 glyphs   [131168 bytes]
FreeSansOblique.ttf: 1333 glyphs   [111580 bytes]
KhmerOS.ttf: 539 glyphs   [265988 bytes]
TSCu_Paranar.ttf: 224 glyphs   [63048 bytes]

Font files: 10
Total size: 1382805 bytes
Unique Glyphs: 2761
Bytes per glyph: 500


removing dejavu (or freefont if the new dejavu can cover everything), would simplify
and reduce overhead (basic latin glyphs, as you mentioned above, are not stripped
to preserve consistency: i.e. at least u0:u7f = 128 glyphs are included in each and
every ttf file including bold/oblique)

ciao,
Davide

Attachment: signature.asc
Description: Digital signature


Reply to: