[PATCH] D113578: [SCEV] Apply loop guards when computing max BTC for arbitrary steps.

[Philip Reames via Phabricator via llvm-commits]

reames added a comment.

In D113578#3195097 <https://reviews.llvm.org/D113578#3195097>, @alexfh wrote:

> (general discussion, not necessarily related to this patch)
>
> (snip)
> Do these sound about right? Am I missing something else? What's your view on the relative importance of the factors above?

You hit a bunch of important points.  Here's a couple extra:

- We want incentives to encourage following upstream development.  We don't want large amount of testing happening far from ToT.  We want to encourage testing as close to ToT as practical because that has the broadest community benefit.
- Some of our contributors are volunteers.  If they spent all of their time fixing issues reported with long delays, they may chose not to contribute.  (This is not abstract - there are areas of LLVM I do not spent my volunteer time on because the overhead of working in that area is too high.)
- Some of our contributors are employees.  Generally, in most organizations getting funding for open ended maintenance is hard.  Having bugs found quickly so that it's within the scope of the current project makes them much more likely to be fixed.  (Ever)

There are many many interacting reasons here.  None are "the reason", all contribute in some way.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D113578/new/

https://reviews.llvm.org/D113578