On Wed, Jul 14, 2010 at 8:15 PM, Otis Gospodnetic
<[email protected]> wrote:
> I'm wondering if the HBase community is interested in replacing the
> Google-powered search (see top-right on all hbase.a.o pages) with
> search-hadoop.com powered search?

I have little experience using search-hadoop.com but from the little
seen, it looks really good.  It has potential of being better than
google search because its tailored to our domain (I like how as you
type, the drop down throws up actual JIRAs, common hbase phrases).

Any chance of indexing commit messages too? (I used search-hadoop to
solve a prob. I was just having -- "When did TestByteBloomFilter.java"
get removed? -- and while search-hadoop turned up the issue that
removed it (the attached patch had mention of
TestByteBloomFilter.java), otherwise, the results were about
BloomFilter and the snippets mentioned BloomFilter rather than
TestByteBloomFilter.java.  Google returned the commit message that
removed the file).

I'd be game for putting it in place if others don't object.  Google is
easy to find if search-hadoop.com is not working for folks.  Downside
is that its my guess that that box gets little traffic.

> And, if so, I'm wondering about the best way to go about it.  I see the HBase
> site is generated by Forrest.  Does that mean that we should check the site 
> out
> of svn, modify the xdocs(?), create the patch, and put it up in HBase JIRA?

Yes -- a patch in JIRA -- though I'd not wish forrest hacking on my
worst enemy. If you can figure it, I'll apply the patch.   How about
hacking TRUNK and mvn site src to?  See src/site.    Here is what the
mvn generated site looks like:
http://hbase.apache.org/docs/r0.89.20100621/ (You get here by clicking
on 'releases' off the main page).   It'll go up when we roll the next
0.89.x release and it'll replace forrest when 0.90.0 goes up -- whiich
should be in the next month or so.

Actually, by default, the mvn stuff doesn't supply a search box so
it'd be a nice addition.

> course, I imagine we'd want to limit the search to HBase by default.
>

I'd say so though fellas tripping on it configured so might miss out
on the nice search of hadoop itself -- if they don't know its
available otherwise (HBasers tend to spend a good bit of time trying
to make sense of hadoop too).

Thanks Otis,
St.Ack

> Thanks, and if you have any suggestions for how to make the best use of this
> search tool for HBase users, please let me know.
>
>
> Otis
> P.S.
> I'll be talking about how we built search-hadoop.com at  the upcoming Lucene
> conference.
> ---
> Sematext :: http://sematext.com/ :: Solr - Lucene - Nutch
> Hadoop ecosystem search :: http://search-hadoop.com/
>
>

Reply via email to