[PATCH] D34444: Teach codegen to work in incremental processing mode.

[John McCall via Phabricator via cfe-commits]

rjmccall added a comment.

In https://reviews.llvm.org/D34444#818047, @v.g.vassilev wrote:

> >> I am open to changing this code as well. That should probably be another review.
> > 
> > I agree.  Are you comfortable with blocking this review until that lands?  It seems like it would significantly change this.
>
> I am afraid that will slow down (if not suspend) our efforts to upstream our local patches. This patch is pretty fundamental for cling and if we change it now, I will have to go back and rework our implementation. I'd be much more comfortable in reworking it once we run on vanilla clang (then efforts seems easier to justify on our end).
>
> It seems to me that despite being suboptimal, it is not very intrusive and it would effect only on our use case. I can keep track of such patches and come back to you for advice how to do them best. Would that make sense?


My concern is that efforts to upstream patches are doomed to get bogged down anyway, and in the meantime we'll have however much more untestable code.  But we have some of that anyway, so it's at least not novel.  I'm willing to accept it.

John.


Repository:
  rL LLVM

https://reviews.llvm.org/D34444