Re: patch for autohinter screwing up rotated text

[Just van Rossum]

At 10:37 AM +0200 02-08-2000, Henrik Gedenryd wrote:
>This is an error very similar to the one I reported, ie. it is caused by the
>autohinter doing a recursive call on code that was apparently written
>without such a use in mind, so the code is run twice. I chose to do the
>bitmap rendering the second time--it wouldn't work the other way around :)
>
>I guess we may well discover more problems like this.

I guess so, too... I suspect the not working FontMatrix might also have to
do with this.

How things currently work with the autohinter is still largely obscure to
me. It would be nice if someone could draw a graph of the dataflow,
involving at least these actions:

- loading a font
- loading an outline from the font
- analyzing the outlines by the autohinter
- gridfitting a glyph
- scaling a glyph to pixel coordinates
- transforming a glyph by a user matrix
- transforming a glyph by the FontMatrix (if any)
- decomposing composites into real outlines

and these components:
- the AH analyzer
- the AH grid fitter
- the various font loaders
- the various glyph loaders
- the face object
- the scan converter

Who does what and when?

Just