[Mlir-commits] [mlir] [vector][mlir] Canonicalize to shape_cast where possible (PR #140583)

llvmlistbot at llvm.org llvmlistbot at llvm.org
Mon Nov 24 14:49:27 PST 2025


MaheshRavishankar wrote:

> All in all, given the nuances discussed, I don’t see a specific blocker preventing this from being merged - or is there?

We might be having different reads of the blocker. To me this discussion is uncovering more reasons why this change shouldnt be merged (this kind of thing is what I was saying would be an issue from the get go). 

> 
> > This does not seem like a great setup.
> 
> I think we may have to agree to disagree here. That said, as `Vector` maintainers, we’re committed to supporting all `Vector` users. If this change ends up causing issues for you, we’ll work with you to address them.

I want to re-iterate : this is not about my use case. We can find ways to work around things either way. So I am disagree-ing more with the approach here, rather than "this doesnt fit my use case".


https://github.com/llvm/llvm-project/pull/140583


More information about the Mlir-commits mailing list