What I've always wanted is a unix command that would give me more info (user generated info) on the error that just happened in my shell.
Scenario: I'm compiling something on Leopard and it stops with an error. I type the command "wtf"
ERROR: compile quit because of this reason xxxxxxxxxx
> wtf
It automatically records the output of the error and saves it some web site's database. It can find others with the problem and open a browser to the discussion on that issue. Someone else probably already found a way around it or there's some library i'm missing and someone there has explained what it is.
The 'wtf' command would automatically do that "google search" for me and log my info to help others.
Heh, that's great to hear. This is exactly what we're trying to do with http://bug.gd and the error sharing database we launched last year. (Along with FF extensions and APIs this year.)
Our long-term plan is to integrate all over the place in software. For example, we have a proof of concept for this we announced at PyCon that does what you're describing inside the Python interactive interpreter:
http://blog.bug.gd/2008/03/29/error_help-for-python-hackers/
While we're excited from a business perspective for integrating into company's GUI apps, as hackers we're very excited about the potential here for completely changing the dynamic of the open source community.
Imagine if every new user, casual hacker, etc., trying out a new open source tool, could immediately get assistance whenever they ran into an error message. What may have taken 6 hours to get configured (and maybe stopped them) could now be a step-by-step "ah-ha" process for common scenarios.
That's precisely what we want to enable. In addition, most of the open source community doesn't even get very good feedback on how many people run into particular errors and give up entirely. This tool would give them the ability to see that, track errors, and put in their own solutions for when they find out people are running into those.
We have a lot of cool plans for this, and I absolutely love to read other people interested in this same endeavor. We're very keen on hearing more feedback on this.
Is this inspired at all by what Fred Brooks talks about in the Mythical man Month, where each run of a program had to be recorded, and was public within the project?
In my experience most of the pay-for-info sites are MS related, or I should say, are for MS-related-technology. I used to find all kinds of great answers in deja and then google groups, but I think that IRC is sucking up all the latest programmer exchanges. Getting all those logs indexed onto the internet would be of great service to mankind.
I think that really depends on how the forum is structured. Usenet suffers from having a horrible signal-to-noise ratio, and while mailing lists aren't bad, they nominally lack googlability and any sort of ranking system.
So, I think the problem isn't that web forums are bad, merely that the interface for using them isn't as good as a newsreader. Might be a good opportunity...
I think that the fact that Web forums don't require readers is what contributes to their popularity. They have a low learning curve.
That said, the fact that they tend to have a government of sorts, can be a big advantage. A forum can potentially do away with trolls and encourage intelligent conversation in a way the Usenet cannot.
If it's going to be a combination of Wikipedia, anti-experts-exchange, and programming reddit, they should have some feature where, when the answer is determined, all the other crap gets deleted, so that people searching into the site to find the answers sees ONLY the answers. (And related tidbits.) Maybe a certain "answered problems format."
Reminds me how when searching for Linux fixes, it is never clear what version they apply for. They might describe some elaborate hacks that have been irrelevant for years, because there are easier ways in newer versions, but it is often impossible to know. I wish I could find only the answers that still apply to my version of Linux.
Someone posted a while ago about a website with a similar idea (I guess it was called bugg'd) where you try to get solutions for your bug, and if you couldn't find in there, you'd be encouraged to post the solution you found elsewhere.
Our services focus solely on error messages-- the goal is to make it so no one ever has to solve any error message someone already bypassed.
While we're a solution database, it's not quite the forum as Joel's site sounds like it will be (more like Experts Exchange).
We use error messages as a vector for connecting users with different solutions. This let's us get a little more exact on the possible solutions, but we wouldn't cover typical programing or sysadmin tips/tricks. For example, we're also a Firefox extension, a Python interactive interpreter plugin, and will eventually grow to capture error messages for all sorts of software and help people right away while they use the applications.
I'll reserve judgement until I can see it in action, but on the surface the idea of a less spammy Expert's Exchange (I think that bit about the javascript windows is a subtle dig at them, but I can't be sure) could potentially be valuable.
Experts Exchange definitely used to do rot13 and for sure do javascript windows. Plus, it's always within a few results on Google. There's no way he can possibly be talking about some other site.
My favorite feature of Experts Exchange? If you scroll down past all the "encoded" comments (which now just say "All comments and solutions are available to Premium Service Members only."), and past the LONG list of topics, what do you find? The unencoded comments with the solutions ;)
There was a time I remember this being true. A few weeks ago I stumbled upon an experts-exchange result and decided to see if it still worked--it didn't. So, I'm guessing this doesnt' always work.
From the horses mouth: "Stackoverflow is sort of like the anti-experts-exchange (minus the nausea-inducing sleaze and quasi-legal search engine gaming) meets wikipedia meets programming reddit"
The idea is, as others mentioned, is not new. Hopefully someone with traction of Spolsky or Atwood will gain enough momentum to make this viable, useful and keep it on top of google results.
Am I the only one who feels that problem they're trying to solve does not exist? Perhaps I read too many books or perhaps my googling skills are bad ass, but finding information on programming issues is not hard and nothing, and I mean it, beats IRC: simple to use, two-way, ad-free, HTML/JavaScript-free, text-only instant and real-time solution to all your problems.
Yet they want to build a giant sink of answers for technical problems. That sink already exists, it's called Internet: post an answer to any question on millions of message boards, google groups ans sites like news.yc. Internet has has an awesome search facility built-in - people will find what you said.
Search for this on google: "The author of this, David Beazley" and you'll get news.yc page with someone's comment containing those words only 4 (!) hours ago (from this moment as I'm writing this).
Why another site? Sure, they have the blogging powers to hype the hell out of their project, but come on...
I think they're going for something a little more definitive. Most of the answers you find online are adequate, but these answers are supposed to be authoritative. If all goes as planned, it would certainly be a benefit to the internet you keep searching on Google, and I think if you're completely honest, you'll remember a few times when you had a really specific problem and you just couldn't find a good enough answer searching Google.
Its a cool enough idea but what's even more interesting is the extent that 2 behemoths of programming blogs being behind something like this will help it be a success.
Exactly. I'm sure Jeff and Joel are both fine programmers, but their audiences, SEO, brands, etc have to be much more valuable. Also, Atwood has shown he can cultivate a large community with his blog comments, and Spolsky has run a pretty good discussion community for years. I think this will turn out well.
(ha - totally unintentional switch from first names to last names)
Scenario: I'm compiling something on Leopard and it stops with an error. I type the command "wtf"
ERROR: compile quit because of this reason xxxxxxxxxx
> wtf
It automatically records the output of the error and saves it some web site's database. It can find others with the problem and open a browser to the discussion on that issue. Someone else probably already found a way around it or there's some library i'm missing and someone there has explained what it is.
The 'wtf' command would automatically do that "google search" for me and log my info to help others.