Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
On Bushido, launch apps from github repos (jacquesmattheij.com)
135 points by jerome_etienne on April 3, 2011 | hide | past | favorite | 43 comments


EDIT: And we're back! The hn-facing server should be able to stand up to a bit more of a beating now, and I've increased the background workers (cheating a bit, since normal bushido apps can't do that yet), so we should be able to tear through the queue much more quickly.

Edit: I've put in a different public server (outside of the normal rotation) for everyone to hit. It's a single box instead of the elastic cloud - so it'll be a bit slower launching for everyone trying it out, but it'll keep the normal sites we run going snappy.

Wow, big thanks to Jacques for such a kind article. He, and several other on hn, have been wonderfully helpful all along the development of Bushido.

If there are any questions feel free to post them here, and I'll do my best to keep up with answering them!

Feel free to shoot me an email at s@bushi.do with any questions or you're looking to get into the beta - be sure to mention hn :)

Oh, and here's a totally gratuitous video for everyone wondering wtf this is, and wondering to themselves, "how can I deploy ~20 apps in 2 minutes?": http://dl.dropbox.com/u/412963/bushido/bushido_hill.mov

I'll start collecting some of the questions here:

Q: It's stuck at 8%! A: Yup. Bushido is self-hosting, because we want to have the same experience as our users. We give everyone 1 background worker for free - and we haven't needed to have more, up until now. We'll add this in later, but for now, just let it hang out for awhile, until the DelayedJob worker goes through the queue and gets to you :)

Q: What happens to my data? A: All the DB data is kept, and backed up (currently it's daily, will go down to hourly), and can be exported any time. Data written to the file system isn't guaranteed to be there, unless it's written to RAILS_ROOT/permanent (or the rails 3 equivalent). If it is written there, it will be backed up on the same schedule as the DB data.

Q: Where is the application running? A: The applications all run on our servers (the ones currently taking a bit of a beating). No need for you to do anything on your side.

Q: What is the end goal of the project? A: Ah, well now, that'd be a bit boring if we just spilled it out right now :) Suffice to say, one-click/automated deployment is just the tip of where we're headed right now. Check out the developer beta if you want to find out more!


Is it just Ruby or can PHP and Python play aswell? Any library/network restrictions? Any tips for making apps deployable?


Just Ruby (rack) and static (CSS/HTML5/JS) apps right now. Our backend is built in a way we think should scale to other languages nicely (although we have to impose some structure), but we're focusing on getting the system and experience right for one group of people first, and once users are sufficiently happy from there, we'll build out in different directions.


Alright, having just deployed several thousand apps with a few more thousand apps in the queue, we're going to lock it down for a bit. We'll be back soon!

If you're interested to get in and try it out for real, let me know at s@bushi.do


As one of the maintainers of a small open source project (http://getontracks.org/), I see a huge opportunity with Bushido. The project has really struggled with making it easy for new users to start up, but even a tutorial doesn't help non-technical users. A few people have set up hosted versions of Tracks, but getting them to stay up-to-date with progress on trunk is a challenge.

With Bushido, projects like Tracks finally have a clear route into the hands of non-technical users. Sure, I can go set up a hosted version of Tracks, but then I'm taking responsibility that, as a part-time maintainer, I have no interest in. This rocks!


I think I was among the first to test http://bushi.do out with http://github.com/SingAlong/billmebob It's a really cool idea for a platform.

This is even better than Heroku. This, if done right, will probably change the way SaaS apps are done. It's the way the web's app store should work. Let people develop apps and let customers host the app, while these two parties never have to care about the hosting for the app (Bushi.do takes care of it. No meddling with config. All done by Bushi.do).


I don't know about others, but I don't think I'd be comfortable outsourcing my infrastructure to someone else to quite this degree. It's sort of like running a restaurant, but instead of having a kitchen, you go to your customers' houses and cook for them.

That being said, this is awesome technology, and it's fantastic for open source projects. I just don't think it will revolutionize delivery of commercial SaaS apps.


It's sort of like running a restaurant, but instead of having a kitchen, you go to your customers' houses and cook for them.

I would say it's like being an expert chef and passing on your recipe to a cook who can make the meal in his kitchen and deliver it to the customer. In this case. You, the developer, are the expert chef, Bushi.do is the cook who does the work for the customers.

IMO, this turns the web like the mobile's app store/market. There's no investment for the developer apart from developing the app (for most apps). He can just put up the git url on Bushi.do. Any customer who's interested in the app can install it on Bushi.do and pay for the resources while you, the chef, gets a cut of it.


It's stuck at 8% (Inspecting url...).

There is no information about the project, what it needs in order to work, what kinds of applications it can run, etc.

I'd be very interested in all of that, I assume hitting the frontpage wasn't too healthy and might have overloaded something, so I'll try again in a few days.


Hey there,

Yes, I put in some rate limiting for the launchers, and it appears to have gotten backed up a bit. I'll see what I can do, but we're in a private beta right now, so I'm more inclined to work on the stuff our regular users are bringing up and worry about hn-scale later :)

That said, shoot me an email, s@bushi.do, and I'll get you into the beta if you'd like, with access to the docs. Though honestly, so far the main support method has been through chat/im/email/skype with me or my cofounder, and we generally take notes from those conversations and turn that into docs.


if so, maybe you can cache the 'top apps', thus people can see what is it about ?


Hi I'm working with sgrove on the project. We had considered an app cacheing or reuse approach for dealing with this influx of traffic, however we wanted every one to have a fresh install experience.

We very well could have directed you to an existing unclaimed install of a popular app and saved the resources of spinning up another. We felt that this could be messy as you might end up on an app that somebody had polluted with test/demo data and settings.

We cache dependencies for popular apps.


i understand your goal. but realize that current experience is "frozen at 8%"... not a super first impression either.

just suggesting to improve the hn post consequences


Same thing happened here. It seems like an interesting idea but there are a whole bunch of questions I was left with, like what happens to my data, where is the application running, what is the end goal of the project, etc. In all fairness I didn't dig around to perhaps this is all explained somewhere on the site.


No no, no need to be fair about that, there's precious little docs publicly right now :) It'll be rectified, but we're deciding which docs to write by working with developers and end users in the private beta.

I'll add your questions http://news.ycombinator.com/item?id=2403003 and answer there - Lemme know if there's anything else you'd like to know!


Can we change the title of the post to something that describes the matter better? "On Bushido, launch apps from github repo's" or something like that.


thanks for the suggestion. done


A comment/question coming from someone who is obviously a designer... It seems like Bushido is operating in a similar space as Heroku, and both have the same purple/dark color scheme and Japanese/Asian theme... Was this intentional, and if so, what are some merits and drawbacks to branding your product similarly to that of a competitor?

Of course this is an early beta and branding is of little importance for this kind of product, but as a designer, it's the first thing that came to mind.


It may be great. But the title should be 'RoR' apps and mentioned earlier in the article. It sounds much more revolutionary until you learn that fact.

All apps are not Ruby or Ruby on Rails!


No, not all; just the good ones :)


And I still think its awesome. I'm trying to try it out right now but seems overloaded. But as an Erlang web developer I feel left out when tools claims to do things like 'deploy almost all apps from github!' :)


You're right, we'll have to work on communication for the public-facing side over the next few days! Thanks for the feedback.

And, I loves me some Erlang, I understand :)


Hah ;)


I didn't expect the word to get out so fast! I've known about Bushi.do for a couple weeks now, and it's really as revolutionary as Jacques says.

And I'm really honored that Sean chose to feature a project I'm working on, LocomotiveCMS, as a way to show Bushi.do's capabilities.


Sacha's an awesome person to work with, for anyone looking for a designer who understands MVP and UX. I've seen him and Didier work magic with LocomotiveCMS!


Ive been following bushido for a while and think it has awesome potential, being able to sell hosted web apps without having to go through the process of building a service infrastructure could bring the benefits of iphone type apps to a much larger audience


I love this idea. Does it only support Rails apps, or does it also support LAMP as well?

I tested it with https://github.com/ginatrapani/thinkup because I'd be very impressed if it could set ThinkUp up with no user interaction, but like many others I'm stuck at 8%.

I'm sure you're getting a lot of traffic from HN now, so I'll wait patiently :).


Just Ruby/Rails/Sinatra for the near future. As much as we'd like to go after that, it's not likely something we'll be able to handle.

ThinkUp looks pretty damn interesting though :)


In the same vein as this, I did a weekend project to write a mac app / chrome extension combo which adds a "Clone Now" button to every github page. It works pretty okay and is available here: https://github.com/patr1ck/Mitosis


Very cool - have you emailed the github guys about this? Tom's insanely focused on the users, and anything that makes github a nicer experience for them. They might blog about it :)


How is it fundamentally different than heroku? (Not trying to troll, genuinely curious.)


For what we're showing today, the main difference is: No account required to launch an app.

We'll be showing more and more over the coming week or two as we start to share developer stories, but I'll give one hint here... http://locomotive-engine.heroku.com/

If you're looking for answers right away or want to try the beta, shoot me an email at s@bushi.do


I know links on jacquesmattheij.com get more upvotes, but can't you just submit http://bushi.do/


Given the amount of information on the bushi.do homepage vs. the description in Jacques' article, I'd wager that would be inappropriate in this case.


That's fair, but I felt that bushi.do would have been equally understood with an appropriate title or comment explaining the website. In fact, if the same article had been written on a different blog, I doubt an indirect link would have been submitted.


perhaps greasemonkey is a good way to get it on the repo page itself


Actually, barring the github guys rushing to integrate with us, we made a bookmarklet I'll post here in a bit.

Great suggestion by the way.


I'd prefer an extension that can be synced between devices and formats/reinstalls.


Nice. Orchestra does something similar for PHP apps.

http://orchestra.io/


its confusing since a lot of github repositories are not ruby, they are not even running on web servers. a bit more info on bushi.do describing what it does would be helpful


You're absolutely right - we also need to explain why apps stick around for 24 hours, how to keep them permanently, etc.

Once the ruckus dies down here later today, we'll circle around to that.


sgrove,etc and github..

Very nice..I like it


Sliced bread is disgusting. I wonder how good this thing is then.




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

Search: