News for hackers. Stuff that matters.
Re:Koders: Search Over 3 Million Lines of Open Source Code A... by Groxx on Sunday November 27, 2011 @01:42PM (#3283019)
Not ready for prime-time. Not even close. I wonder if they ever tested their query engine, it's hardly useful. No quotes? Queries ignore important characters like "_"? No roll-up of identical results? It also misses methods in its "(num) methods" results (or throws them into a totally different order) in several languages, and has some crazy match-ordering behavior.

An example: I searched for "helper_method" (a Rails method), and found these (bracketed what was bolded as the search hits):

  class.Scheduler.php:
  // Holds [method]s for ...
  Editor.py:
  from Models import Editor[Helper], Controllers
  AutShare.h:
  // IShare [method]s
and then I get over a dozen literally-identical files called "helper_method_example.rb" (tests with these results:

  require File.dirname(__FILE__) + '/spec_[helper]'
  
  describe "a context with [helper] a [method]" do
    def [helper_method]
  
  and the rest of the files contained this after those lines:
      "received call"
    end
    
    it "should make that method available to specs" do
      helper_method.should == "received call"
    end
  end
There are about three pages of results like that, or the same wrapped in a `module HelperMethodExample`, with identical contents otherwise.
Re:Koders: Search Over 3 Million Lines of Open Source Code A... by alextgordon on Sunday November 27, 2011 @02:08PM (#3283102)
It's been like that for years. Can't see it improving, now that they're going to have a monopoly on the market.

I will happily pay money to anybody who can build a code search service that is actually functional.

Re:Koders: Search Over 3 Million Lines of Open Source Code A... by dncrane on Sunday November 27, 2011 @02:52PM (#3283208)
This isn't exactly the same thing, but I launched a search engine for programmers (that is, it doesn't ignore special characters) a few weeks ago: http://symbolhound.com/

It doesn't search code repositories, but rather tutorials and question/answer sites like stackoverflow. Right now stackoverflow dominates the results but that should change as we index more sites.

Re:Koders: Search Over 3 Million Lines of Open Source Code A... by boyter on Sunday November 27, 2011 @06:05PM (#3283674)
Would you be interested in paying per search, paying for access or just general donations based on your experience?

I ask as I am doing this but have not considered a business model for it as yet.

Re:Koders: Search Over 3 Million Lines of Open Source Code A... by Groxx on Sunday November 27, 2011 @07:00PM (#3283791)
Micropayments suck, currently. If that ever changes, I'd be in favor of both per-search and access; until then, maybe a freemium setup? It'd be hard to hook programmers without letting them try it, and you can experiment with more computationally-expensive searches (regex, AST, etc) without letting random users tank your system.

I wonder if you could even build a market off open-sourcing the engine, thus allowing people to improve it for you, and run a paid service that searches GitHub, Google, etc for results. Free accounts for contributors that fix bugs or provide speed / feature improvements?

I dunno, just throwing ideas out while they're percolating, I haven't given them a whole lot of thought.

Re:Koders: Search Over 3 Million Lines of Open Source Code A... by boyter on Sunday November 27, 2011 @07:32PM (#3283881)
Yes I don't like micro-payments myself either. Freemium is my current plan.

I had also considered the GitHub model. I will be making portions free-software as the whole thing is built off it anyway.

If nothing else I will be going free for quite a while as I believe that there should be a code search repository available online.

For the record it will live at http://searchco.de/ I was in the middle of pushing out my second release when Google made its announcement so its looking a little worn at the moment.