[Mlir-commits] [mlir] [mlir][python] add dict-style to IR attributes (PR #163200)

[Maksim Levental]

makslevental wrote:

> So I'm asking why aren't exposing the new API instead and applying the great improvements in look&feel to the APIs that we want user to actually use.

The answer has already been provided to you: because *this* patch is wholly orthogonal to that work. Let me put it in terms of a very simple metaphor: if the foundation of my house needs replacing, I can still fix the lock on my door (to replace a faulty mechanism, to prevent thieves, to make ingress/egress easier for laborers working on the foundation, etc) *even if the the new foundation will require eventually a new door frame*. 

> However the fact that we still have in-tree users for the old API does not change the fact that we need to promote and encourage APIs...

I don't know about others in this PR but I certainly do not feel *encouraged* when someone shows up and *demands* people interrupt their work to satisfy arbitrary, unrelated goals. In fact, I feel extremely *discouraged* about my (and others') ability to make meaningful progress on work without constantly having to satisfy someone's demands. Furthermore, as we always say: "patches always welcome". If you have been dissatisfied for an entire 2 years with the relative lag of the bindings behind core (in our disuse of the new APIs), you should have felt at liberty to send PRs rectifying both our disuse! And your own dissatisfaction! In fact you should feel free even now, today, right this moment to send any such PRs!

> which will make users of this API less happy, which I suspect is something we all strive to avoid (unnecessarily at least).

If your actual concern is the user facing API, then I feel you do not understand patch itself (as I often feel at these impasses) - **this PR does not in any way change the actual user facing API**. What it does is complete the currently incomplete implementation of the [`dict`](https://docs.python.org/3/library/stdtypes.html#mapping-types-dict). In whatever future you envision, this is the contract you will have to fulfill, *irrespective of whether the current contract is unfulfilled*. By blocking this PR, all you are doing is gating the correct implementation of that contract behind properties, which is a complete non-sequitur.

Thus, I am once again asking you, politely, to remove the block, and move this discussion elsewhere (e.g., a discourse).

https://github.com/llvm/llvm-project/pull/163200