Show HN: I've open sourced Bookmarkly, a bookmarking site written in JavaScript | Hacker News Reader

Show HN: I've open sourced Bookmarkly, a bookmarking site written in JavaScript

github.com

70 pointsdangrossman17 comments

This was a weekend-ish project to learn Backbone.js, among other things. http://bookmarkly.com is running on a small linode so it's a little behind on generating screenshots after some large imports.

You have your password/session salt checked into github, you might want to remove that :)

Checking out site now. First thing I've noticed is that the Chrome plugin doesn't work. Keeps referring to localhost.

http://localhost:3000/bookmarklet/?url=...

I thought I had changed that before posting this. I've updated it in the Chrome directory again. Thanks for letting me know.

What are you using to generate the screenshots? I see a link to http://screenshots.bookmarkly.com/ in the code?

Edit - just seen this:

> The screenshot generating server is currently a separate repository that I have not yet released.

Any plans to do this? Been looking for a decent screenshot solution for a while now...

I also wrote a screenshot server (in JavaScript) for the project which is running at that subdomain. It's using Webkit for the screenshots then node-imagemagick to crop and resize. I hope to open source that part eventually too, but I have a bunch of things to fix still (some bugs in my worker/queue system, and some problems with my setup of the X server), and I need to remember all the steps I took to get it working to document it for others.

Unfortunately AFAIK all the current tools for automating webkit/khtml still require an X server so it's not quite as easy to set up as cloning a repo and running the code.

http://url2png.com/ would make a good drop-in replacement in the meantime, but it's not free.

I'd love to help out if you are able to open source it one day. I spoke to the owner of url2png and he's a great guy, but I'd rather roll my own services for that sort of thing!

It seems like getting a working setup with headless X is the tricky part...

Why not use something like Sauce Labs? I built a screenshotting service on top of them using soda (https://github.com/learnboost/soda) a long time ago, and the code's bit-rotted by now, but it was quick and easy, and combined with firefox's full-page-screenshot, you can get some awesome results.

Their entry-level plan is free, so you could easily have a "put your Sauce Labs credentials in this file and we'll take screenshots in the background".

I've had success with wkhtmltoimage for this, it's part of wkhtmltopdf: http://code.google.com/p/wkhtmltopdf/

phantom.js can do this, and the bonus is that it's already written in javascript. :) It depends on QT for it's webkit component but that's it... Doing a screenshot with it is pretty much a one liner:

page.open(address, function() { page.render(path); });

http://www.phantomjs.org/

Does it need to be an online service? The Selenium2 webdriver <insert-language-here> bindings make this really easy and it already supports firefox/ie/webkit/iphone/android etc (http://seleniumhq.org/docs/04_webdriver_advanced.html#taking...)

Not interested in hosting your own image screenshots? Check out http://linkpeek.com

That was the original plan, but your site charged me then errored out without giving me an account when I tried to sign up. I hope you've fixed it (and tested it) since then.

The site doesn't work on IE8. The signup form sends a GET request. This is where it redirects me to: http://bookmarkly.com/?email=sample@email.com&password=1...

BTW, I don't normally use IE. I just happened to today.

It's been fixed, at least in IE8 mode of IE9, thanks.

I'd switch from Diigo, if we could save page caches.

Many others being held hostage because we can't export our page caches would encourage you.

Screenshots is a good start.

Great project!

One bit of feedback: I wish clicking the bookmarklet would keep me on the page I'm bookmarking instead of taking me to bookmarkly.

Reply on news.ycombinator.com