The unofficial Google shell | Hacker News Reader

The unofficial Google shell

goosh.org

396 pointsJCB_K50 comments

That's cool, but I like the Google Command Line interface even more:

http://code.google.com/p/googlecl/

I was blown away the first time I edited Google doc using vim, hit ":wq" (save), and saw my changes reflected instantly in my browser, where I had opened the same document.

The Oauth authentication was also pretty cool. Instead of asking for your Google credentials every time / saving them on your machine, the first GoogleCL use opens a browser window, prompts you to log into your Google Account, and asks you to authorize the Google CL client. You may then revoke this permission at any time (like any Oauth token).

I played with googlecl for a bit and quite amazed by what it can do. However, in order to use vim as the editor, the user has to always append '--editor vim' to the command line, which feels a little awkward to me. I want more than that, say, it would be great that those google docs are actually files!

A quick google search turns out there exists always a project doing that --- http://gdatafs.sourceforge.net/

"Gdatafs is a FUSE implemtation that mount picassa web and youtube to your filesystem. An experimental implmentation of google docs is also avaiable in trunk. "

I think you can specify a default editor in the googlecl config file: http://code.google.com/p/googlecl/wiki/ConfigurationOptions

You can also specify document-type specific editors ;)

The beauty of googlecl is that, since it's a command line tool, you could always fall back on using a bash alias / bash function, if all else fails.

Would certainly be nice if it listened to the EDITOR environment variable for a default :/

Actually, now that I'm at home, I just tried it... and it does default to the EDITOR variable. Just tried it by setting the variable to vim and then to nano.

Hi, I made this some years ago. In fact it was the reason to sign up as a hacker news user :)

I didn't had the time to update it, so some things are broken (login, auto complete). And the open source project never caught traction.

Yeah, I notice

su -;rm -rf /

doesn't delete all the data from Google's datacenters...

(It was worth a try, right?)

I fixed this after the first time ;)

You might be a UNIX geek if:

You type "cat news" and then are confused that the results involve felines.

or you fat finger "cat mews"

I'm just shocked that something that uses a Google search API survived for more than 2 years.

http://news.ycombinator.com/item?id=1025039

I take your repost and match it!

http://news.ycombinator.com/item?id=206683

Wow, we're forgetful, aren't we.

Here's how I see reposts. I agree that people shouldn't just repost things because they were popular at one point, and they want the karma. However, realize that for people who weren't around when that was posted (over a year ago), this is pretty cool. There is also no way I would have ever thought "Hey, I wonder if there is a shell-esque interface to google? Oh, I bet HN will have a link about it! Let me go search!". So, I think when spaced out enough reposts are good. Over a year seems like long enough to me.

I disagree. I want to see new stuff. There's plenty of worthy stuff that in /new that does't see the homepage because shit like this reappears every 6 months/1 year/whatever.

edit: emphasized what kind of stuff in /new

Well new stuff to you might be old stuff to me. Unless you've looked at every single submission on this site, no need to complain. Just don't open the link or comments.

It's really hard to use the site with that mindset when 15/30 of the submissions on the front page are resubmissions. I'm not sure if that ratio has ever happened, but it certainly feels like it quite often.

Other articles that make it feel that way that I wish I could just eliminate all together--anything talking about Node.js, productivity related (how I fooled myself into working), "how can I learn to program." These all seem way too common for my tastes.

I'm not against any of these topics, in moderation, but regardless of what anyone says, the community that has grown here over the past years has become dominated by those with poor taste.

You are projecting your own preferences. You don't like Node.js stories, I'm sure an equal number of hn users prefer those. I also highly doubt that half of the home page are resubmissions, based on the fact that a lot cover recent news (see the Apple iPad coverage for instance).

So either come up with your own filters, but don't try to push them on everyone else here. We are diverse enough to enjoy a bit of everything.

No way! I wrote an opinionated comment? How rude of me!

> So either come up with your own filters, but don't try to push them on everyone else here. We are diverse enough to enjoy a bit of everything.

Are you suggesting that if I build filters I not share them with the community?

I think the point I'm trying to make is that Hacker News is no longer one size fits all. If I have to spend my time here searching amongst the crap to find something I want to read, it's just flat out not worth it. I've already started going back to Reddit much more often, because of their subreddits (after years of maybe once every 6 months or so browsing).

HN knows what I've clicked when logged in, seems like the sort of thing that could be used to hide resubmissions if desired ...

That's a good idea. It could be an option in your account preferences.

I think "shit like this" is fairly subjective, so you can see the rest of the thread for that part. I agree with MikeCampo, this was certainly not old to me, and I truthfully enjoyed it and would have never even found it without the repost.

Apparently you like re-flamewars, too. Can't we leave this kind of stuff for meta posts?

Pretty Awesome. I'd love it even more, if it had auto-complete on the command-line too.

Looking at the svn repo (http://goosh.googlecode.com) this project is over 2 years old.

Or to be more precise, over 2 years not updated.

I've made something like this for DuckDuckGo, but haven't released it since I want tab-completion (uses a static list for now). Will release as an official UI once DDG gets search suggestions.

http://dhruvbird.com/tty.html

Edit: There is no GA tracking on this page even though I use to for my site (other pages) - to keep with the general privacy experience of DDG

I love the use of the url fragment.

Type some commands and hit the back button a few times.

Very cool. `man login` says that you only see the username we type in. How are you authenticating?

I'm guessing it all happens client side from the browser.

I don't want to "guess" that. I'd like it really spelled out. I tried "login" and got a 500 from a debian server. I wouldn't send my credentials through some random server just because it was on HN.

--edit: I was wrong, it wasn't a 500 error, it was a 404. I don't want to cast too much suspicion here, just making a point about being appropriately paranoid about giving out my creds. This site could be totally cool, I want it transparent to me before I login.

I'm assuming it's trying to load an OAuth screen - but the page is broken so won't even work to begin with.

While the client side has some login-related logic, it looks like authentication happens server-side.

From the source:

  goosh.module.login = function () {
    this.name = "login";
    this.aliases = new Array("login");
    this.help = "login with your google account *";
    this.helptext = "goosh sees only your username";
    this.parameters = "";
    this.cb = function (A, B) {
      if (goosh.ajax.iscontext(A)) {
        if (B && B != "guest") {
          goosh.lib.cookie.set("loggedin", "1", 365)
        }
        goosh.config.user = B;
        if (goosh.config.user != "guest") {
          goosh.gui.outln("You're logged in as: "
            + goosh.config.user + "<br/>");
          goosh.set.init();
          goosh.gui.updateprompt();
          goosh.gui.showinput();
          goosh.gui.focusinput();
          goosh.gui.scroll()
        } else {
          goosh.ajax.query("http://goosh.appspot.com/dologin?")
        }
        goosh.gui.showinput();
        goosh.gui.focusinput();
        goosh.gui.scroll()
      }
    };
    this.call = function (A) {
      goosh.ajax.query("http://goosh.appspot.com/
        status?callback=goosh.modobj.login.cb")
    }
  };
  goosh.modules.register("login");

Wow I've been thinking of writing something very similar lately using nodejs and parsing the html (maybe using lynx --dump). But this is pretty much as good idea and blazing fast!

Like this? https://github.com/fictivekin/webshell

Appeals to the inner command line in me. Needs a spell command to make it complete for me :)

I enjoyed this quite a bit. Set it as my new homepage.

If it's unofficial its 'An' instead of 'The'

Not if there's only one unofficial entry in this field. "The" entails that the following noun phrase uniquely identifies something, not that it's been endorsed by some entity.

guest@goosh.org:/web> calculate 1 inch in m Error: could not calculate that.

:-( fail. But still a really great concept!

cool !

typed pwd got search results.

Confirmed. Same here. Also, options for ls don't seem implemented.

cool

goosh goosh goosh...

Fun

Gah. "sudo make me a sandwich" returns actual results.

There has to be a few easer eggs.

'su root' doesn't seem to work. It just shows search results. I'll file a bug report.

Reply on news.ycombinator.com