@lassiecoder@expo@amanhimself thanks for getting this added 🙏
we did try to make this smoother (forcing a different path limit across cmake) but there were so many small blockers, so great to see the messy rest documented cleanly 🙌
@Apex0646@expo you'll have to check the native logs for that as Expo Go can't surface those. The "Viewing native logs" section here has more info https://t.co/NANOZVJjES
@diip3sh@expo it's a maintenance and alignment question, mostly. We could write a plugin for it, or integrate it if we open up a transform plugin API. however, there's several ways forward that we'd align on first via a RN working group with the React Foundation
We've released noxcturnal with Expo SDK 58, and it's open source now. It's an oxc-based transformer for Metro with a JS visitor layer. You can opt in with "experiments.noxcturnalTransformWorker".
On the T3 Code app's codebase, it transforms modules 4.3x faster than Babel
@ericsonl_@expo@t3dotcodes the goal is full runtime equivalence to our Babel preset so hopefully it'll never be all that scary 🫡
we'll also figure out settings to tweak some mode switching (hybrid Babel for app, noxcturnal for node modules for example)
@vanilagy@Doctorthe113 it sometimes hangs for longer, although that's gotten rarer. The problem is w/o batch release support you either get misaligned publishes or with staged releases you're doing one-by-one click-ops
https://t.co/p77xNKBL6U
Measured on the same 4,076 modules, median of 7 runs with a cold Metro cache on an M3 Ultra. A full iOS dev bundle goes from 4.13s to 2.15s. Worklets and RN codegen still go through Babel (367 files) and now take up over half the remaining transform time
@tchayen@t3dotcodes t3 connect basically just works™️ so given the pairing flow, grok can then use the MCP endpoint ✨grok bot is an amazing meta agent / coordinator, I've found
Spent all week with 37 Claude agents grinding in parallel, 18 hours a day, until they fully reverse engineered @expo.
sharing the results so you don’t have to burn the tokens. 100% free and open-source:
https://t.co/bePSdnvdHg
@ergis_mullai@expo@krystofwoldrich Device Hub support and other changes were backported. Generally, unless there's conflicts or risk of confusion all fixes or changes like these are backported to prior SDKs within the supported range
We just shipped an updated T3 Code Stable release with over 1,400 changes 🤯
Quick highlight reel of my favorites:
- Massive perf wins on desktop and mobile
- Add a project to another machine from the "Run on" menu
- EAS Simulator remote hosts for iOS dev
- PR diffs now show code vs tests vs docs at a glance
- Idle Claude threads compact early so the cache stays warm (beta)
- Mobile got a big native pass. It finally feels native
- `t3 doctor` can reclaim database space
- `t3 connect` now checks your machine is reachable, and errors say how to fix them
- Webhook tasks got filters and dedupe
- Skill discovery is much more reliable
- Tons of Windows, terminal, and scroll fixes
Least favorite: I caved, LaTeX renders on web and mobile
Also heads up, we're getting a bit more experimental with how we treat Nightly, so now would be a good time to move back to stable :)
@CxnVI25dKV96358 the ngrok one was restricted in capacity and at some point anti virus started flagging it, since shipping a tunnel is deemed suspicious. That then eventually made its way into Windows Defender's flagged binaries
@CxnVI25dKV96358 (sorry for replying in english 🙏)
Expo SDK 58 has a new tunnel that replaces the ngrok one. I know maybe the upgrade isn't on everyone's todo list yet, but I'm hoping it'll be a better experience
@LittleBrainz@expo@WojciechDrozdz0 yes, this is (currently) specifically built for Apple's App Intents (so not to be confused with Android intents or expo-sharing's support for sharing intents)
@sanketsahu@Baconbrix the last part of cold builds is now held back by the transformer. Eventually, we'll figure out plugin APIs and locking Metro configs down to move to noxcturnal by default
the bottleneck with it enabled now though are modules that fall back to Babel
@sanketsahu@Baconbrix the last SDK alleviated some resolver bottlenecks (some do remain though and this also is currently not a task that can "run ahead" of transforms yet by speculatively parsing dependencies). that's on the main thread
it also worked a bit on the in-memory FS layer