Shouldn't be long now.
A huge amount of work has gone into making FFMPEG and x264 as fast as they are today. Not only would you lose performance (and integer types) going from C -> JavaScript, but you also lose all the assembly, handwritten for tens of different architectures (including SIMD like SSE). I would expect at least a 100x slowdown.
For your users' sake, please don't do this ;_;
FFMPEG...maybe not. But there is certainly room for something cross platform that works for upload videos quickly.
Anyhow, great job doing it. Depending of the pricing, I'd surely use it.
I've seen that you make your widget code open source. I guess this handle the flash hack / html5 version for all different versions of browser? Hopefully so because that's the most annoying part of getting the ajax upload working. But then, if that's free, I'd probably take these widgets and just manually code the amazon part.. The only reason I'd use the service is if the pricing is a very little more than what amazon would cost me.
However, that's just for me. Other startups or bigger company might be willing to pay the big price it can save them hours of coding.
And, to everyone you've never done the ajax upload.. trust me that it's a pain. So many browsers and hack.. compressing the images, storing them, limiting the size, etc etc etc. It seems like a trivial task and the first thing you know is that it's been a week you're working on that and then, once you think everything is done, you receive weird message like it doesn't work on IE9 for some reasons, or whatever.
we don't need anything back, because we generate uuid on the client
Wrote a small service in Haskell to handle large uploads in preparation for a new feature. Does interleaved IO and streams the uploaded file directly to Cloudfiles while also saving a copy on the local disk to be able to extract some metadata and add them to our DB at the end.
Also did play with the idea of creating a general upload service based on this implementation, but for now I'm focusing on other things.
Let's say I use html/css from scratch; or integrate some coffeescript in it.. How clients are supposed to edit the text? I.e How the intuitive widgets works to change stuff?
Webpop lets you work with structured content, pull the content into your templates and define exactly the HTML used to present the content.
The system then automatically detects what part of your site is dynamic content that the client should be able to edit and what's design. The client then get a simple on-site editor for updating his contents.
There's a video of it all in action here:
http://www.webpop.com/blog/videos/on-site-editing-the-easies...
Your site was nicely readable until the custom font got loaded. It is tiny and a colorful blurry mess (subpixel) on my monitor. The text right above the bottom image is almost unreadable.
-When I upload files, are you serving them infinitely, or just caching them so I can fetch them from my server?
-Are there any x-domain issues here? Can it work over SSL?
2. We haven't tested it properly yet. But there are no rocket science; I believe, everything is going to be fine as soon as we get the SSL certificate
(Also, by "illegal content", think "Feds literally busting down your door and hauling you away" sort of illegal content, and by "you", I mean, you, va1en0k. This is not necessarily a reason to stop right now, but don't get too far down this path without legal consultation about how you should protect yourself from this eventuality. There are things you can and should do, like make sure you qualify for DMCA safe harbor provisions.)
However, in my dream, the service becomes so awesome so it inspires new, better standards! :-)
For my particular application, I'm looking for something which can handle large files (500MB - 2GB), preferably giving file upload progress and being able to resume broken uploads. Could you talk more about those features?
Also, it seems like we would upload files to your servers, and then to integrate them with an app, I would then use your API to bring them over to my servers or storage. Are there plans to make the source of the uploading widgets/mechanisms themselves (not necessarily open-source) so that we can host our uploads ourselves?
Ooh, and how about something that is mountable?
Anyway, those are just a few of the things that immediately crossed my mind in terms of use-cases for my current project. I can see this being very nifty!
Our widgets are open source: https://github.com/uploadcare/careful-uploaders
To handle videos, I'd recommend you to check out http://transloadit.com/ . They are something like our competitors, but I believe we have different targeted audience.
We've been thinking about splitting files in parts and everything, but not right now.
Thank you for your suggestions!
You may want to take a look at how Twilio handles in-browser audio recording (even though they offer something different). Basically, you insert some JS into your page, then the audio is streamed to Twilio, and then Twilio does a callback to an endpoint you specify (with meta data, the URL to the file etc etc). What is also great about this is that I can serve the files right of their servers and not have to worry about it as a developer.
Best of luck with it!
