On 2011-03-29 3:26, Josh Andler wrote:
On Mon, Mar 28, 2011 at 6:18 PM, Jason Creighton <jcreigh@...400...> wrote:
On 03/28/2011 02:54 PM, Alexandre Prokoudine wrote:
>
> Instead how about better SVG compliance? It's not that fancy, but
> still very important for the project. Or animation.

Hmm, do you think there's enough compliance issues for an entire GSoC?

Way more than one student can handle in one summer project. So, on that note, it's a GREAT area to choose from. Tavmjong, would you mind chiming in on some of the more "important" compliance issues that you think may be viable to be part of GSoC work?
There are probably enough "random" SVG compliance issues to last you a summer, but two of the specific areas that could use some love are:
 - Markers (especially their orientation)
 - Support for operating in the linearRGB color space (this basically affects EVERY filter we render, so it's not unimportant to fix asap).
There are probably a few more (Tavmjong or ~suv will probably have a few more suggestions), but if you want to have a closer look yourself, there is a rendering test suite... (The wiki has instructions on how to get it and use it.) I would recommend anyone working on our compliance to update this to reflect changes in the W3C test suite and to use it to gauge their progress (this would of course be a perfectly valid part of such a GSoC project). It would even better if you improved the rendering test suite and/or made it more accessible.

Also, although not strictly compliance, we do have a long standing issue with paths that needs a solution and could use some good thinking. Basically the problem is that in SVG the "last" segment of a closed path HAS to be a line segment, but that we do NOT want to show expose this to the user (at least not always). So technically we should always show nodes for this last (often zero-length) segment, but this feels very unnatural, so we try to avoid that at all costs. The straightforward fix is to pretend the segment isn't there if it's zero length, but this gets problematic real soon when transformations and/or relative coordinates come into play. (I've suggested to the W3C that they fix this in SVG 2, as they're redesigning the path syntax anyway, but that'll probably take a decade or so.) So, can you think of a better solution? (This project might not be sufficient on its own, but could very well be part of a larger GSoC project.)