Rafael de Oleza

JS at @stripe

[Rafael de Oleza @rafeca]

@emilsjolander unfortunately I don't know a better way to debug these issues :( in our case the issues came from OSX showing an auth popup which is very hard to debug without having remote desktop access

[Rafael de Oleza @rafeca]

@emilsjolander I'm glad it helped, did you move away from storing the certificate in base64? That's a pretty neat solution and I'm considering doing the same (we're currently downloading it from a private GitHub repo)

[Rafael de Oleza @rafeca]

@ladybenko - big shops usually 8pm, smaller local shops 6-7pm
- modulor is a must for arts and crafts!
- the barn! you can buy them on any of their cafés

have fun in Berlin!!

[Rafael de Oleza @rafeca]

@kuvos @mjesun @github Code freeze!! 🤷‍♂️😅

[Rafael de Oleza @rafeca]

@fatih At Facebook most of the codebase is in the monorepo, which doesn’t have ACLs or any ownership rule (anybody can modify any file), but this requires a very open culture with lots of trust in the engineers

[Rafael de Oleza @rafeca]

@_zamotany @dan_abramov @syke @callstackio @haul Yes you're right, in order to make Metro inline your imports you need your Babel config to leave the imports intact 😅 (it's just that Metro internally uses a Babel plugin to do that inlining), but yeah that's part of the bundling process.

[Rafael de Oleza @rafeca]

@_zamotany @dan_abramov @syke @callstackio @haul Still, that's just some babel plugins that interact well with the bundling logic of Metro, so it should be possible to port this to WebPack and other bundlers

[Rafael de Oleza @rafeca]

@_zamotany @dan_abramov @syke @callstackio @haul Yes! actually, we initially tried to use babel's lazy mode but we saw performance regressions (caused by the additional logic that the lazy mode forces to execute very time a require is used). After this, we decided to implement the inlining of import statements inside Metro

[Rafael de Oleza @rafeca]

@_zamotany @dan_abramov @syke @callstackio @haul RAM bundling is a different concept than inline requires (although RAM bundling extremely benefits from inline requires). As far as I know, WebPack/Haul do not support inlining the require/import statements

[React Native @reactnative]

Six months ago the community told us Hot Reloading was a major pain point. A better Fast Refresh mode is on it's way into React Native. It's resilient to errors (no need for full reloads after a typo) and supports function components with Hooks. Check it out:

[Rafael de Oleza @rafeca]

@sebmck @mjesun Could you use this step to encode a list of imports/exports to later do dead code elimitation of unused exports?

[Rafael de Oleza @rafeca]

@cnakazawa @sebmck @mjesun Well it would also get stripped with inlineRequires enabled, since then modules are considered to be without side effects

[Rafael de Oleza @rafeca]

@cnakazawa @sebmck @mjesun On the TypeScript case since it’s a normal import it would not get stripped (unless the babel plugin to strip typescript type definitions also removes the import, which idk if it does)

[Rafael de Oleza @rafeca]

@sebmck @mjesun Makes sense, relying on the side effects of a module that was imported just for its types is quite an edge case 😅 i was just curious if Rome did some kind of pre-analysis of dependencies before building the graph

[Rafael de Oleza @rafeca]

@madbyk @fb_engineering Thanks man! I can tell you that the changes are noticeable! 😁

[Rafael de Oleza @rafeca]

@void_0 @MikeMonteithDev @cnakazawa 😂😂

[Rafael de Oleza @rafeca]

How to get your name on almost every blame of a project 🙃

[Rafael de Oleza @rafeca]

This is an example of an existing PR where we fixed the git conflicts: https://t.co/UFrrRh88Ve

[Rafael de Oleza @rafeca]

In order to not cause inconvenience to other contributors, I created a (very hacky) semi-automated script to fix the conflicts caused by this commit on existing PRs and I ran it against all the opened PRs that had been updated in the last 6 months.

https://t.co/zyKX778fMu

[Rafael de Oleza @rafeca]

@marklangovoi Thanks for the suggestion! While rewriting the history may work in some situations, this was done on a big opensource project with almost 8 years of git history, >36K commits and thousands of forks. In this scenario, rewriting would have caused high disruption to the contributors