29 comments

[ 25.7 ms ] story [ 81.1 ms ] thread
In case this gets broken anytime soon, I see:

Repositories: 4,236,263

Code: 962,748,183

Issues: 5,952,195

Users: 4,472,663

At this time (4:48 AM EST, 2013-10-11), minutes after Hacker News changed the parent comment's timestamp from "1 hour ago" to "2 hours ago", these are the updated values:

Repositories: 4,236,957

Code: 962,598,538

Issues: 5,953,084

Users: 4,473,412

So in a space of time probably equal to either 1.5 hours or 2 hours, the numbers changed by these amounts:

Repositories: +694

Code: -149,645

Issues: +889

Users: +749

Repositories: 4,237,352

Code: 962,617,133

Issues: 5,953,624

Users: 4,473,778

Repositories: 4,237,870


Code: 962,651,473


Issues: 5,954,482


Users: 4,474,320

It seems to work with @{any letter}. Can anybody explain to me how GitHub is interpreting this search term?
(comment deleted)
Sorting by "Least recently joined" is interesting. The GitHub team are all there, naturally, but I'm intrigued what "tater" is about. Yehuda Katz is famously user ID 4 but tater shows up as user 611789 through the API.

Other than that, the first 10 pages or so are a real who's who of the Ruby scene in 2007-2008 :-)

(comment deleted)
Woop. 140th person to join GitHub.

Back then you pretty much had to be invited by a Ruby/Rails person. I got in by submitting a patch to Dr Nic. :)

See, I'm not entirely sure about this date ordering stuff.. because I'm around #80 by that method, but if I use the API, I'm #118. Try http://caius.github.io/github_id/ to see what your actual user ID is, as I think it's more truthful.
So now you could scrape the name of every GitHub user and their email if it is public.
Oh they rate limit and only allow access to the first 100 pages of results. Well that is that idea out of the window :P You could probably get around 7000 of them max, barely a scratch.
It'd take you about 48 days :) When I checked that ID of the most recent user was 5663608. The github API will give you 5000 hits an hour if you authenticate your requests. You'd have a load of "not founds" but working up from 1 to 5663608 would get you every public user (no private accounts or banned / deleted).

5663608 / 5000 = just under 48 days worth of non stop API harrasment... maybe that explains why github keeps going down, someone is already trying this out :P

Even if it's not :)
You could also get away with a search for created:>2007 Not really sure what the "@" is doing.
Assume this is going to disappear once they wake up in the US...
If I search "@z" I get 13 repos. So why exactly do you think that searching "@a" is yielding all repos?
So how does this work? From the search its cheat sheet:

"@defunkt Get all repositories from the user defunkt."

Seeing as there _is_ a user called "a: (http://github.com/a) but he doesn't have a single line of code, no repositories and no active issues either; the search breaks somehow and returns the repository, issue and code (LOC?) count of ALL people.

For example trying it with "@b" (or any existing user after the at-sign) does yield the correct results (the respective counts for user b: http://github.com/b). Trying it with someone with no repositories, code and issues OR a user which does not exist (@thisuserdoesnotexist) results in the same behavior.

@a is probably an alias for @all (https://github.com/search?q=%40all&type=Users&ref=searchresu...).

[edit 1] @everybody seems to return every user

[edit 2] @everything seems to return ... everything.

[edit 3] @qdfhsdfjsdqjrekle seems to return everything. Ok. It's a bug. :>

No, they are specifically searching for content by @{username} - refer to the cheat sheet on the search page. @{non-existent-username-or-user-with-no-activity} triggers this behavior.
This query is incorrect, because it doesn't include private repos and forks. The headlines on this page https://github.com/search are accurate. As of this comment.

Search more than 4.3M Users

Search more than 8.8M Repositories

Search more than 18.7M Issues