Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Arnor — a marketplace for Django projects (arnor.io)
86 points by lukeman on Jan 3, 2012 | hide | past | favorite | 12 comments


This looks perfect. I am currently developing a self-contained open-source Django app I aim to allow people to deploy on as many PaaS as possible, but I was bummed out that I still had to include a minor guide to the necessary git and terminal commands.

One-click deployment for Django has been a long time coming from bushi.do (http://www.gobushido.com/), so it's great that you are pioneering the Django effort.

Now I just hope that the prices will be low and the tiers intelligible to people with little to no proficiency in web programming, when even I struggle to get the concept of dynos.

PaaS -> One-click deployment is without exaggerating the thing I am most excited about from a technological perspective because of the implication for people unfamiliar with the requisite technology.


I'm a little confused. As opposed to creating a service with user registration, and maintaining the uptime of this service, instead, I create a Django project that someone can buy? And all maintenance on that instance is handled by gondor.io and that user? If this is correct, then who is my target audience as a developer selling on Arnor? Tech savvy people, only other Django devs, or the general masses that would want to use my code as a service? If it's general masses, how have you made the barrier to entry very low? I don't want to explain what an "instance of my code" is to my mother, but I would love to sell her a django project having to do with cute cat photos :)


Here is what was said about Bushido, a similar service for Rails that currently is based on one-click deployment of GitHub repos:

https://news.ycombinator.com/item?id=2402894

https://news.ycombinator.com/item?id=2438002

https://news.ycombinator.com/item?id=2939115

Remember to read the actual articles.

Obviously, the "Django" part makes absolutely no sense to a lot of people, so you don't have to market it like that.

Think of it as a Mac App store for web services, where you don't have to download and compile the binaries, get the dependencies, etc. We are already seeing a revolution in desktop installations with Steam for Windows and Mac, where all you do now is order a game which is then installed - I doubt people today even know there is something called DirectX.

The current way to install internet services for "normal" people is to download a project (from a download URL, not something as abstruse as GitHub where you can struggle to just find the friggin' download link), register a domain, pay an internet host, set up the MySQL database in your cPanel (which looks like shit, in case any of you haven't used it recently), after which you spend half an hour bugging your techie friends about what the hell a public key is, and how the hell you convert the thingamajig to .pkk - this is all after which you have tried finding out how the hell you can connect to your server, so you can drag and drop all the downloaded files into a window. On Windows, this can make a grown man cry, especially when you are using a cheap-ish host and haven't ever heard a whisper of ssh.

When Platform as a Service came about, layers like database, application, caching, load-balancing were abstracted and delegated, but you still had/have to get the latest version of a project or repo, get the dependencies/CLI, and push it to the server. This involves git, the terminal, pip (easy_install/apt-get). After this, you need to configure your service, which is probably done with environment variables and can not necessarily be prompted as a native interface outside of the shell or terminal. Even so, it's a much better experience, because it means you have to concern yourself with a much narrower knowledge domain, because so much of the stack is abstracted, but there is still a lot to be done. With one-click deploys and such, this may no longer be a problem (and any configuration can be presented in a nice interface, because the gondor guys only need to concern themselves with one particular stack, where this most likely will be set up with environment variables in Python).

You might still have a slightly hard time seeing what this exactly means, seeing that there aren't a lot of services to give you an idea, but that's because this obstacle have held back a lot of ideas and projects, and I suspect that we will see a completely new type of services that we couldn't imagine before, because it wasn't possible. We can't extrapolate in the same way from what we are used to, because that would tantamount to imagining that the results of the combustion engine are horses with engines instead of cars.

Just imagine that we might be able to deploy an entire web service from an iPad without any proficiency with Bash, ssh nor programming.

I guess some hackers will never see this, because they want to mess with the gritty innards of everything, so it is possible that this appeals more to people with iPhones and Macs than those using Androids and Ubuntu. The services are most likely as-is, but when people are paid to maintain and update them, users might be able to create services that "just work". And I think the fact that it's a marketplace where people can be paid for the apps mean that we get some very important incentives to create - and maintain! - quality services. Think Apple's App Store.


I'm pretty dense, but who is creating what for whom here? I create a Django app on Arnor then sell it to someone who wants to maintain it? A client logs on to Arnor and requests an app and some developer builds it for them?


Whereas Gondor is for hosting your own code; Arnor is for hosting instances of other people's code. So it's a marketplace for Django developers to sell (or offer for free) out-of-the-box Django projects that people can deploy with a single click, customizing to the extent allowed by the original project creator.


Can someone (or you guys) take this further and create a marketplace for developers to sell, rev-share, or license fully developed applications to marketers, individuals or companies?


That's the core of what Arnor will be.


I don't usually submit "joke" comments to HN, so I'll resist the urge, even though there is ample opportunity atm.

Great name. In fact, probably the best name.

The idea sounds interesting. I wonder what the benefits might be of having a marketplace on top of a PaaS. I wonder if there are any actual benefits of that.


Tell me if I am understanding this:

I can create an application, like a website for realtors (as a simple example), and then customers (like realtors) can pay me and get a hosted instance of my web app. Right?

If so, cool!


Hm, we're using gondor.io for an open source project, not sure what is implied by "... We've built the hosting and deployment infrastructure known as Gondor. Now all our efforts are coming together as Arnor."

I presume Gondor.io will continue to exist as it does now?


Absolutely!

Think of Arnor as a layer on top for people that want to use other people's code for their site.

Arnor itself just deploys to Gondor and you'll always be able to deploy your own code to Gondor completely independently of Arnor.


My guess is that Arnor will be tightly integrated into Gondor. One-click install of apps similar to Heroku.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: