On Sun, 2004-04-04 at 15:42, Bryce Harrington wrote:
Are we using Pango through some of the other libraries?
Well, Gtk obviously. I don't think it's really even relevent to any of the other libraries we use.
What sort of approach do we want to take for doing this changeover?
Is the Pango API parallel enough to libnrtype that we could just do function exchanges, or will it require more significant wrapperizing or rewriting? Is this the sort of work we could divide up amongst us all, or is it more of a 1-2 person tiger team type change?
Not entirely sure yet. I know it's different enough from libnrtype to do simple replacements, and we will need to provide some equivalent of the current libnrtype <-> libnr interface ourselves (notably, the mechanism for getting glyph outlines into the renderer).
Probably the best sequence to attack this in might be:
- write a new libnrtype backend using pango
- remove all the other libnrtype backends
- remove the extra level of abstraction for the pluggable libnrtype backends
- peel away the libnrtype facade from Pango where we can and start using the Pango APIs directly
- refactor the remaining libnrtype bits into a few adaptor objects/functions to get glyph outlines and text attributes to and from Pango
- start emplying Pango to perform text layout for e.g. SPText and friends
If someone is feeling particularly motivated, we could probably short-circuit a couple of these steps.
-mental