[llvm] [mlir] [MLIR] Add debug log to the pass manager (NFC) (PR #156205)

[Andrzej Warzyński via llvm-commits]

banach-space wrote:

>  I've been tweaking this with Jacques for the last month :)

Yes, I've noticed and really appreciate all the improvements. Also happy I finally have cycles to track this a bit more :)

> > Btw, a nit, [please avoid force-pushing](https://llvm.org/docs/GitHub.html#rebasing-pull-requests-and-force-pushes):
> 
> I'm not sure how to avoid this actually: any chain of work I have locally is unmanageable for me without this (one PR == one commit)

Sure, I know where you are coming from and it makes sense. However, from what I can tell, it's a "chain of work" downstream (as in, I don't see any other patches upstream except for https://github.com/llvm/llvm-project/pull/157194). So, from the "upstream" perspective (i.e., the perspective that reviewers like me have access to), this is an isolated change. 

By force-pushing, we are loosing the history. Specifically, it is impossible to "connect" comments to code updates and, effectively, to understand how this change evolved (and the rationale underlying the design). All in all, it's a tradeoff - the development is much easier, but that comes at a cost of lost history.

I am merely sharing my understanding of our development policies and the rationale behind them - I've been making similar request to other contributors. If you feel that I am misinterpreting our policies, we should probably start a discussion on Discourse and clarify the docs.

On a related note, with https://github.com/llvm/llvm-project/pull/156205 we are effectively creating "stacked PRs". So, IIUC, this [commit](https://github.com/llvm/llvm-project/pull/156205/commits/7606fa3d0b689abbfbd6446dc9f1019e38567106) no longer belongs to this PR. Could you either update the summary or use one of the other methods that we use for stacked PRs? 
* https://llvm.org/docs/GitHub.html#stacked-pull-requests

Thanks Mehdi!

https://github.com/llvm/llvm-project/pull/156205