Fork me on GitHub
#fulcro
<
2024-01-30
>
Ruslan13:01:14

Hello! 🙂 Is it worth learning Pathom 2 or better learn Pathom 3 and start use it in production?

Eric Dvorsak14:01:02

They are both pretty similar from a user perspective, so it doesn't matter too much if you start with 2 then moves to 3 at some point. I found the best thing about pathom3 is how good it is at batching, but I was also quite disappointed by how slow it is when you have a lot of resolvers (it takes 500ms to compute the plan, and I had issues where a wrong plan is cached and then it sticks to it, with caching it obviously goes does to almost 0 after warming up), and how slow it is when you return a LOT of items (count more or less 1ms per item so +1k items takes more than 1 sec where a hand built SQL response with multiple query would take under 100ms)

tony.kay17:01:00

I’m still using 2, and I kind of prefer how it does async handling, and I have a ton of code written around it, but I am constantly tempted by the better batching of 3. Interesting to hear about the planner problems. That kind of thing is always tough to get right.

genekim00:01:58

I recently learned that Fulcro Inspect resolver index doesn’t work with Pathom3 — @U0522TWDA has a tutorial of how to replicate it, but i didn’t have time to try to make this work. (Reminder to self: write up the problem I had to Jakub) In that moment, I definitely wondered whether Pathom 2 might be more desirable, as it seems to be the more traveled path, and has the best support inside of Fulcro and RAD.

tony.kay02:01:37

Yes, there’s that. The fact that I don’t yet use 3 means the tooling isn’t going to get attn too quickly, unless others step up to do it