RPC transition

6 views
Skip to first unread message

Sean Colsen

unread,
Apr 2, 2024, 2:47:37 PMApr 2
to Mathesar Developers

Team,

I wrote a wiki page to lay out the plans for our RPC API transition.

There are not many hard decisions or design choices in this document yet. Rather it’s a sort of scaffold at the moment which serves to map out the shape of this transition process and give us a place to fill in decisions as we make them. That said, there are a few decisions that Brent and I made last week, for example “always used named parameters”.

Pavish, I think it would be good for you to read this document in full and make sure you don’t have any objections to any of theses plans.

Kriti, you might be interested to skim this document to get a rough sense of this plan since it is currently blocking the beta release.

After we get clarity on special characters in method names, then we can begin filling out our intended function names. I would imagine using the PR process for that so that others can weigh in. We need to get agreement on the patterns we’ll be using before we run wild with implementation.

Brent, once you figure out if we can use special characters in method names, I’d be happy to take a stab at opening a PR that proposes answers to a bunch of our open questions and our function names. Then we could open that PR up for review from other team members in order to move the process along. (Or if you want to open that sort of PR that’s also fine with me.)

Kriti Godey

unread,
Apr 2, 2024, 11:14:12 PMApr 2
to Sean Colsen, Mathesar Developers
Thanks for laying the action items out so clearly, Sean. I have skimmed it and have no comments at this time, it looks good. I imagine we'll discuss it a bit in tomorrow's weekly meeting since we're planning out the release.

Pavish Kumar Ramani Gopal

unread,
Apr 4, 2024, 3:55:59 AMApr 4
to Sean Colsen, Mathesar Developers, Kriti Godey
I read through the document in full and it looks great and well laid out!

The questions/action items seem to cover everything we'd need. I do not have any objections.

For `id`, I'm strongly against sequential ids. I'm good with fixed ids which is also the conclusion you're leaning towards. When/if we adopt websockets, we would have to move towards UUIDs, and I don't think we should care about that now as mentioned in the doc.

I'm in agreement with only using named parameters in requests.

Everything else seems good.

I would like to read through the doc again once we have decisions for all the items.
Reply all
Reply to author
Forward
0 new messages