Address comments

Addrwss comments

Thanks for the comments. I addressed them in https://reviews.llvm.org/D146456.

Could you share a reference use of these interfaces? I'm not very familiar with this myself, just remember that it broke our tests last time.

.

.

Address comments

mlir-print-ir-after=pass-name is easy enough to use from the opt tools. However, I could not find an easy way to set that programmatically (or pass the flag through the tests)

Rebase

Fix formatting

Rebase

Fix flang

Rebase

Rebase

Fix Cmake file

Thanks! Default values are possible but then it collides with the other builder

Address comments

Thanks!

Address comment

Thanks!

.

Address comments

Add TODO

In strong favour of this.
All IR should be readable, especially linalg, which is not the easiest dialect.

In D141928#4059151, @jpienaar wrote:

Could you perhaps expand on that exception or point to where this is assumed? Are these where during a pass an invalid state is created or is this across passes?

We cannot generally infer the return type as there are use cases relying on the old behaviour.
In particular, cases which expect an empty yield op to be generated just to replace it afterwards.

Don't feel strong about this though, so we can discard this also.

Right, getValueOrCreateConstantIndexOp works.
As there is a builder for int64_t, which creates a constant one for OpFoldResult would be the natural completion, no?

This failed our internal tests. I think the issue is *not* with your revision but a test case would be nice here!
Could you add a test for the reshape op explicitly?

I am not conviced that this is an improvement but this is highly subjective ofc.

The Ci is running but I do not see any changed. This is strange.