Wikia Acquires Grub, Releases it Under Open Source

An anonymous reader writes "During a keynote address at the O'Reilly Open Source Conference (OSCON), Jimmy Wales announced that Wikia has acquired Grub, the original visionary distributed search project, from LookSmart and released it under an open source license for the first time in four years. Grub operates under a model of users donating their personal computing resources towards a common goal, and is available for download and testing."

So?

[DogDude]

It's a nice idea, but badly implemented. Search doesn't even work (I've tried it before, and it's never worked), and on top of that, today, search throws a bunch of mySQL errors (mySQL? for a search engine?!?!). It's a nice idea, and maybe this group will breathe some life into it, but right now, grub is doa, as far as I'm concerned.

\ I'll be excited if they make a working search engine. I'll be even more excited if they do work on the searching algorithms, which is what makes or breaks search engines, not the amount of content they have. It's what you do with the info. Anybody can spider lots and lots of information. Providing relevant results is why Google is #1 now, and those results are the results of some heavy, heavy math that is done by lots of pHD's. Spidering, which seems to be the focus of Grub, isn't a big limitation for most search engines these days.

Is this a good thing?

[Animats]

Wikia is a for-profit company. Users running portions of their crawler should be paid. At least in stock of the company. Otherwise it's a ripoff. It's reminiscent of Kazaa's approach to "peer to peer": user machines do the work; Kazaa collects the money.

Distributing the web crawl isn't that big a win. The crawl is a batch job, but replying to search requests is a near real time application. The expensive part of a search engine is the system that generates fast search responses. That's where you need the systems with gigabytes of RAM and tight coupling to the other machines of the cluster.

Doing the web crawl on user machines offloads some of the effort, but not all that much of it. If you want to cut crawl costs, some of the query machines can be devoted to crawling during slow periods.

Remember, you can't trust the client. Web spammers can modify their copies of the crawler to report extra, phony links to their web sites and boost their stats. This gives a whole new meaning to the term "link farming". Until Wikia, there was no easy way for "search engine optimization" types to mess with the internals of the search engine. Now there is.

Besides, what's the selling point? "Our search costs less to use than Google?" Hello?

Re:FIST SPORT

[Afrosheen]

*That* doesn't suck, GRUB does. I have no idea why Debian distros have such a hardon for GRUB.

Hell, the fact that it doesn't have an uninstaller (like LILO does) should be reason enough to avoid it. lilo -u and it's gone. Grub? Hope you have a dos or other bootdisk with fdisk on it handy.

Re:FIST SPORT

[VagaStorm]

I think what your looking for is your boot disk and the command grub-install --some_parameter_to_to_set_root_directory :p

Re:FIST SPORT

[caluml]

I also had a problem with Grub that I spent about 2 days trying to fix. I can't remember now exactly what it was, but it was something to do with a raid controller, Reiserfs/XFS, and something else.
And when you're building a server, you can't sit searching the web all day.
Lilo works. I'm not going back to Grub.