SVG backend doesn't fill loops if they occur in the same subtree as a line

[byorgey]

It looks like #42 was a bit more subtle than I thought. See the last line of http://projects.haskell.org/diagrams/backend-tests/all-index.html .

The problem is that a style node may occur at the root of a subtree which contains both lines and loops. In that case it seems that the only possible solution is to actually split the style node and push fill attributes down until they are over a subtree which contains only loops.

In fact, perhaps we even ought to do this as a compilation pass, in diagrams-core. We could make sure that fill attributes are only ever at the root of subtrees containing all loops. Then backends wouldn't have to worry about it at all, but could simply issue a fill attribute when they see it.

[byorgey]

To do this in diagrams-core would require adding Typeable constraints to primitives, so we can examine them without giving a Renderable instance. We've decided to do this anyway, so I'm not worried about that, but in any case this is not a quick fix. I'm inclined to leave this until after the 1.0 release --- this is not the sort of thing users will be running into a lot, and if they do, there are simple enough workarounds for the moment.

[jeffreyrosenbluth]

I agree, we should not try to rush a fix in before 1.0. Additionally, I
don't think 1.1 will be far behind.

On Thu, Nov 21, 2013 at 4:17 PM, Brent Yorgey notifications@github.comwrote:

To do this in diagrams-core would require adding Typeable constraints to
primitives, so we can examine them without giving a Renderable instance.
We've decided to do this anyway, so I'm not worried about that, but in any
case this is not a quick fix. I'm inclined to leave this until after the
1.0 release --- this is not the sort of thing users will be running into a
lot, and if they do, there are simple enough workarounds for the moment.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/43#issuecomment-29025369
.