Coming back to meteor :-)

very old meteor fanboy here, guess even my forum account here is kinda ancient, and just getting into what has changed these years. Also the guy who wrote this thing The Renaissance of Meteor.js | Anonyfox quite some time back. and now here we are, and I want to start giving back things to the community as I encounter friction along the way of building stuff.

first thing dropped today: GitHub - Anonyfox/meteor-publication-collector && The trusted source for JavaScript packages, Meteor.js resources and tools | Atmosphere … and I try to get the hang of “properly” doing typescript packages for atmosphere. please let me know if it works if you need it. and yes, its inspired from much older community packages that somehow either are dead or no longer working properly with meteor 3+ async stuff, so expect more to come that also might involve porting older ideas into something working again :slight_smile:

long story short, just wanted to say “hi!” guys :slight_smile:

13 Likes

Hey man, glad to see old faces coming back to Meteor.

I’m interested to know what sparked this act? Was it a spur of the moment act or did you learn something and wanted to see the latest changes?

Also in case you didn’t know Meteor is about to drop a change even greater than fibers-to-async migration

yeah I know its coming but also I am now a bit careful about timelines-til-release after having followed meteor dev progress on the sidelines for a while. essentially I have to ship some production grade stuff right now that ideally works for a while and also right now, so I don’t actually want to mess with RC candidates currently.

and here I have some quite complicated publication logic I wanted to unittest and the preexisting packages on atmosphere didn’t work well (or not anymore) so I whipped up what I needed and how I’d expect it to work, especially with a focus in speed (I have lots of tests) and really don’t want to use brittle timeouts that also slow down things needlessly. digged into the meteor source with some LLM help and stitched together a sharp focussed solution, and decided to share it. and I guess I have more things I can soon extract and publish.

what made me spark in general? well last time the hard constraint on the old node version was actually a final dealbreaker for some important projects, now I saw the latest progress being made resolved most of my issues (so i am unblocked) and also I made the react-router-ssr thing work again on react 18 at least (non-negotiable for my projects, direct SSR is the dealbreaker for me if not working proper), and I have some super time-constrained projects and decided to look into my (decades old!) secret tech weapon for truly fast building&shipping. and last but not least: galaxy now has direct mongodb hosting, so I really can ship stuff there directly. Money/Costs is not my concern here anymore, raw speed and convenience is, even though I have years of devops knowledge and could configure stuff manually, I cannot afford even a single day of wiring shit together on infra sometimes, and multipe SaaS vendors for single app feels just wrong. Now galaxy looks like its one stop shop (well maybe some S3 bucket in the future needed but hey), including github integration, and from my testing (even though its beta still) the core seems to work well enough.

so yes, meteor DX/DDP + React/SSR + Accounts/* + CI/CD on click + unified direct hosting = exactly what I need and want. For actual feature building, Cursor+LLMs is already a thing, Meteor solves the backbone that holds the features for me perfectly fine.

3 Likes

heyyy @anonyfox! Great to see you here :slight_smile:

We named The Renaissance of Meteor.js event after reading your article last year. I guess we talked about it, right?

Anyway, welcome back!

3 Likes

Hello @anonyfox, nice to see you here!

Yeah, beyond rspack for release 3.4 taht will bring an insane performance for apps build process, we are cooking something for the ObserverDriver layer for release 3.5 (PR), we will add changeStreams as a reactivity option and deprecate oplog (some day)

3 Likes