I found this report, do you mean that?
http://projects.blender.org/tracker/?func=detail&atid=498&aid=32588&group_id=9

The report was closed because Brecht thought he fixed it. If you don't agree with that, a single reply to a report will always be handled in a fair way.
For work on our tracker we expect from our reporters a pro-active attitude. We work together on an honest attempt to make Blender better!

Let's ask Brecht too.

The seam in normal map baking I don't see in my build, attached a .blend file that's setup to just press Bake where it bakes ok here.

Which build are you using? This fix is not yet in an official release, but should be in the daily builds here:
http://builder.blender.org/download/

The texture baking issue I'm not sure about, the attached .blend file doesn't have any textures mapped, and I don't see the issue attaching a texture myself.

The normal-map is OK. But the texture has the same problem (s. attachment).
Are you using x^2 textures for your tests?

Sorry. I add the packed blend file. And I have adjusted the view.

The texture baking problem happens because of a tiny mismatch between the low and high poly models. The low poly outer boundary vertices are located at -2.0 +2.0, while the high poly ones are located at -1.99981 +1.99981. It's made worse by the uv going exactly through the middle of the pixel here, snapping the UV's to the pixel grid avoids the issue as well.

Maybe there's an epsilon that can be added somewhere in the code to avoid this, but then it reduces accuracy in other cases. Not sure yet how/if this issue should be solved.

Fix in svn with an epsilon, since the difference here is so small that it might have just been float precision while editing the mesh that caused this.