Re: [cp-patches] [generics] RFC: Genericize return type in PrinterStateReasons

[Tom Tromey]

>>>>> "Andrew" == Andrew John Hughes <address@hidden> writes:

Andrew> Generally, I think these kind of edits are okay, but I'd save the
Andrew> majority of it, where possible, to after the generics branch is
Andrew> re-merged.

Generally speaking, outside of java.util I've been pretty much doing
a shallow genericization -- only the API and not the contents.  There
are a few exceptions to this, but not many.

My thinking here is, a shallow one makes merges simpler, and we can
fix up all the warnings when we're closer to closing down the branch.

However, as you are doing the merges, feel free to set the policy here
to what is convenient for you.

Andrew> Additionally, it
Andrew> could possibly catch the odd bug... though I've not seen it yet.

There was one bug on the trunk that was caught by inspection, but
which would have been caught by genericization.  I forget what it was
exactly... something in ClassLoader or System or something like that.

There were a number of javax.imageio bugs that were caught by
genericization.

I think that might be all I've found.  I suspect that when we do the
deeper genericization we may find a couple more.

Tom