[Mlir-commits] [mlir] [MLIR][Tensor] Canonicalize fully covering slice insertions into tensors with unit prefixes (PR #92912)

[Mehdi Amini]

joker-eph wrote:

>  Bottom line, if I look at the commit history and project affiliations of the author and the person rebutting, I'm willing to believe they both have an intuition and feel for the algebra here. 

It is perfectly fine (and actually I would say desirable) to have "intuition" from experience, however it does not free oneself from being able to **articulate** some reasoning, which is entirely absent from the discussion here.

In particular the lack of first principles reasoning (including trying to anchor some of it on "experience" and trying to make a "feel for the algebra" more concrete) is quite alarming to me. 
We are in the simplest possible situation here: these are two operations from the same dialect, operating at the same level of abstraction, and the semantics of the two forms seems completely equivalent (any of these statement may be wrong, but I have seen a rebuttal in the thread!). We're pretty far from anything "excessive' here!


https://github.com/llvm/llvm-project/pull/92912