Hacker News Reader

Comment by zongitsrinzler | Hacker News Reader

The real takeaway from this is that the author uses Hackathon Starter (https://github.com/sahat/hackathon-starter).

I have used it for multiple projects and it gives a huge head start compared to starting from zero. Signing up, logging in, resetting the password, uploading, etc all seem like easy work but when you pile them all up you can easily spend a week just getting to the point where you are within minutes of cloning the starter repository.

However the failure of GoChat is not relevant to Pokemon Go. While GoChat might have done something very wrong comparing 1mil users to an app with tens of millions of concurrent users is invalid. Pokemon Go would be a NoGo running on a single Node.js machine without any sort of balancing.

View on HN

> However the failure of GoChat is not relevant to Pokemon Go.

That part is almost right, but also a little confused.

He built GoSnaps, which is a geo-enabled gallery with: No chat, few messages, few writers, many readers, medium amount of reads, large messages, not time-dependent. (Not even user state.)

Then he contrasts that to the failure of GoChat, which is a geo-enabled chat with: Many-to-many chat, many messages, many writers, many readers, many read operations, small messages, very time-dependent.

The performance profile between these two things is wildly different and the way he's patting himself on the back over what amounts to half knowledge is pretty disgusting.

There is a real feeling of superiority in his post based on his own assumptions (in terms of the other apps stats). I always think it's best to constructivly critique something by not trying to promoting yourself in doing so.

what if he just didn't mention gochat? its an awesome post

Failure of GoChat? GoChat didn't fail, it just had some hurdles it needed to jump. GoChat is back up 100% baby! (I'm the guy who started GoChat). I reached out to the author of this story and he has since updated his story. I now have a team of 6 dedicated guys helping me full time on this and we're cruising forward.

That is great to hear. Good work! Sorry i can't edit the post anymore, otherwise i'd update that. :)

Edit: Wow, he did update it and he's only defending his bullshit with more half-knowledge. At this point i'm ready to call him a liar, just over this claim: "My conclusion is that both apps are very similar in terms of scalability complexity."

Really cool. Do you know a list of similar repos for other stacks? C#+JS? Java+JS? ...etc

Php?

Try JHipster for a Java (Spring Boot + Angular) stack: http://jhipster.github.io/.

Django : https://github.com/DrkSephy/django-hackathon-starter

Anyone knows the same thing for elixir/phoenix?

Yes, it's a pain and particularly the user authentication solutions are very complex. The best I have found is this example but it's 100% not a starting point as the first thing you'll have to do is rename everything!

https://github.com/hassox/phoenix_guardian

For a second I thought that was a plugin before I realized it was a sample application.

Yes, I think the Phoenix guys would rather you forwarded to an OTP app than build a mechanism for plugins.

I think no-one builds their apps this way yet so maybe it needs more explanation.

I find this thread a little confusing. That sample application is using Ueberauth (https://github.com/ueberauth) which appears to be a rich plugin-based authentication framework with many preexisting plugins. You don't run the sample application, you use the framework.

The point is it's not fantastically quick to drop in, in the example app look at the UserFromAuth helper - it's definitely not simple and contains a load of gotchas I'd struggle to deal with from the ueberauth documentation.

The grand parent was asking about a template/hackable starter kit with batteries included. That setup of ueberauth is the best I've found sadly.

>Yes, it's a pain and particularly the user authentication solutions are very complex. https://github.com/hassox/phoenix_guardian

This is why I don't see the point of moving over to Elixir. Rails has ready-to-go and battle tested "modules" that you can pop in and go.

If I used Elixir now, I would have to wait for the community to build something battle tested or I can home-roll my own Elixir authentication. The Poster above just said, "It's a pain particularly user authentication."

It's definitely a lot harder to move to Elixir and Phoenix but I've enjoyed learning FP and Elixir a lot; I've done it mostly to challenge myself and it's informed my programming in other languages.

I maintain that a web framework is probably one of the least interesting things that's enabled by a language like Elixir, a VM like BEAM, and a foundation like OTP... but... for whatever reason web frameworks have become the criteria by which we seem to measure languages these days.

That said. A chat application is really, really well-suited to Elixir.

Really? The user authentication in the book Programming Elixir was pretty simple and straight forward—is that not "production ready"? Authorization seems like a simple plug away.

Or any python-based solution?

https://github.com/pydanny/cookiecutter-django is fairly similar in terms of providing a good jump start for django development.

A quick thanks for that one!

Any flask based ones?

https://github.com/mjhea0/flask-boilerplate might fit the bill, not as comprehensive though.

Very true, even though not speaking specifically about the hackathon-starter but for any similar boilerplate project. If you enjoy prototyping and making MVPs, having a boilerplate that you are personally comfortable with and which covers most of the functionality is pretty much no-brainer.

I've been considering making something like this for myself. I love my current stack: TornadoWeb with a Gremlin Server graph database backend, but I do find myself recreating it from time to tim. Maybe I should put the effort in to at least skeleton the thing/define common functionality so that I can quickly prototype.

Reply on news.ycombinator.com