Drop UMD bundle from distribution #191
Conversation
There was a problem hiding this comment.
Thanks, I know it's draft and will keep it open. But I believe dropping UMD and moving away from rollup are two different discussions. I understand the benefit of tsup, and will happily discuss this as well with the team, but as mention do know that we extensively use rollup in all our SDKs, and I dont see any strict need to drop rollup here that justifies the unalignment or the introduction of alignment work in all of our other SDKs.
Additionally, the npm run dev command does not work the same anymore, so we will need to bring that back if we were to drop rollup. But again, different conversation as it impacts our general tooling we use with our JavaScript libs/SDKs.
No need to make any changes here, I reached out internally and will get back to this PR.
|
Have had an internal conversation and we might want to see if #192 would be sufficient before considering a new bundler. Given the size of this library, we might not even need a bundler but the question is if this would cause any issues that TSUP would solve. Will need to continue investigating this, but wanted to open a draft PR to allow for a conversation. Long story short, we are okay to drop UMD but let's figure out the bundler. |
|
I also dropped UMD on its own here: #193 |
|
I think it would be nice to have as little tooling as possible, so I certainly agree that the best option would be to simply rely on the TypeScript compiler 👍 |
This is a draft PR to show what it could look like if we were to drop the UMD bundle (see #190). Specifically, it would allow the elimination of a lot of dependencies (including Rollup), by switching fully to tsup.