[prev in list] [next in list] [prev in thread] [next in thread] 

List:       graphviz-interest
Subject:    [graphviz-interest] Graphviz Charset Issue (Pixelglow UTF-8-ness)
From:       erg () research ! att ! com (Emden R !  Gansner)
Date:       2004-06-30 10:45:03
Message-ID: 40E2D1EC.6000509 () research ! att ! com
[Download RAW message or body]

I agree that there is a problem. The phrase "doesn't care" about the charset isn't quite
precise. It was meant as an intended policy and was meant to apply to the user. That is, the
user should be able to provide labels using the ascii charset, an extended ascii charset
(codes < 256), or the UTF-8 encoding of Unicode, as well as a font, and get the correct output.
Certainly, some internal part of graphviz needs to worry about the encoding.

The first case is easy. The last two cases cause the problems. Postscript works fine with
extended ascii and bitmap output (thanks to libgd) works fine with UTF-8 and, at least up
to now, extended ascii. (I'm hoping you are just seeing a bug in your PNG output; libgd
apparently decides what encoding to use based on the font.) The main open problem is how
to extend postscript output to Unicode.

Some of us would allow just UTF-8 and assume the user will convert the graph using some external tool.
Based on the "doesn't care" policy, I would like us to continue to support extended ascii. If we can't
or won't make the decision automatically, this make require the user to set an attribute. (This problem
has also arisen in the context of SVG output.)

As Glen notes, this is a "transition period".

     Emden

[prev in list] [next in list] [prev in thread] [next in thread]