Thinking about this more, I think some kind of more structured, declarative security, i.e. Roles and ACL, is an important feature. Allow and deny rules don’t seem enough for real-world scenarios.
Parse has this built in, and also uses MongoDB. Parse is now open source, so maybe some of these security features could be adapted for Meteor?
If I had to name one thing about Meteor that has always been a pain in the ass for me, I’d say it’s that bloody awful Atmosphere.
I’m not sure if MDG guys are taking it into consideration, but for a casual candidate for a Meteor developer, Atmosphere is one of the first real Meteor webapps they ever see and usually the first they actually use. And the experience of using it is terrible.
It’s just soooo sloooow, both with access to data and with its shiny UI. And it leaves an impression of “what? Will my real life Meteor projects be so shitty too?”
What Meteor is missing is a complete rewrite of Atmosphere or redirecting people to Fastosphere if Atmosphere can’t be rewritten.
But Fastosphere may be gone too, with *.meteor.com going down soon.
You know, I don’t really use atmosphere much as I generally know what packages I need. I’ve found it to be a clunky experience so I just don’t bother, but until you said this, I never put thought into why I just don’t bother. You nailed it exactly.
In Meteor 1.3 npm packages will work out of the box. From there it’s only a few steps to having Meteor packages live on npm as well, so eventually there will be no more need for Atmosphere.
I want to hear a beep when meteor compiler finds an error and, therefore, doesn’t reload my page. As it is now I wait for a page refresh, sometimes it’s coming sometimes I wait too long to understand that I have to tab to the console window to check what error there is. Isn’t it a low hanging fruit for improving dev experience?
There are plenty of great community packages for these. Core doesn’t have to support every possible configuration. In fact, what they’re trying to do is get less and coffeescript out of core.
Two different issues here
The first is solved by flow router, and again, it doesn’t need to be core, the entire eco system is defined as meteor.
The second would be awesome!
You should make a request for that on github. If they approve it and are willing to accept a pull request, I will look into adding this.
Any chance you can get something Apple in there? And IBM? I’m suddenly very curious if someone can come up with an application from every huge name in computing! (Include any I missed too)
Visual IDE E.g. where you just add Template component and can create events (onCreated, onRendered, onDestroyed) just by double-click. Something from Visual Studio and Delphi world but really convenient. Though it is probably huge amount of work to make it work both ways (code <=> ide).
This is something Geoff talked about during his ‘bad idea file’ on Transmission, as in… it would be really cool to do, but would be a time-sink from a business perspective, and is still years away.
I’m not so sure. Personally, I think we may be less than a year or so away from an initial release. We have many of the pieces in place with the Atom editor already. And since Atom is already built on Chrome, the heavy lifting of getting the rendering engines aligned is already done.
The next step is to create a style sheet GUI sidebar for Atom, much like the Visual Studio ‘Properties’ pane. Once that package is written, we’ll have the pattern for GUI creation of components, and Events/Controllers tab for JS will follow. Last part will need some drag-and-drop functionality for the HTML.
The biggest technical caveats is that Atom is written in CoffeeScript. So, my prediction is that the first Visual IDE released into the community is going to be based on Atom, using a Coffescript component system. And what view layer supports Coffescript the best? Blaze. And who is working on CoffeeScript components? Mitar. (As I understand)