Re: [ft-devel] Experimental: v38 interpreter with minimal backwards compatibility mode and linear advance widths

[Nikolaus Waxweiler]

So I've been running the code for one week or something with only minor 
changes here and there and I think it's in a good enough shape to ask 
for comments. Werner? ;) Also, what about Infinality's code? I'd be 
eager to remove it completely, but obviously the decision is up to you.

Known deficits:
- Can't switch between v35 and v38 in ftview, the program appears to 
hang. Haven't investigated yet.
- There is at least one (web) font (Essay Text) that manages to change 
advance widths, even though it shouldn't. Must still track down.
- No explicit handling of known tricky fonts, those should probably get 
the v35 interpreter unless they happen to be native ClearType fonts.
- Still haven't checked if I can short out more things, would need to 
fiddle with FontForge so I can get a graphical debugger. Fonts like 
Times New Roman, Palatino and Georgia might benefit some more.
- Must still implement additional GETINFO flags.

A note on advance widths: I know that bytecode can change it around, but 
would that be compatible with FreeType's fractional advance widths 
support thing? My v38 always declares that subpixel positioning should 
be used and prevents most changes, read: is focused on linear advance 
widths. Given that hinting on the x-axis is mostly prevented and I see 
little use for changed advance widths outside of a Windows environment, 
I'd opine that v38 should do linear advance widths only.

experimental-v38_v4.patch
Description: Text Data