search engine in apache docs

It would be nice to replace the current hacked-up google
search of our
docs with a proper "custom search engine" from
google. The problem is
that custom search engines are tied to specific google
accounts, and I
don't think it is a good idea for any single person to have
control
over the config.

What if we created a google account for the docs project and
stored
its userid and password in the private pmc svn and used that
to create
the search engine? Does this violate google's terms of
service?
Opinions?

Joshua.

------------------------------------------------------------
---------
To unsubscribe, e-mail: docs-unsubscribehttpd.apache.org
For additional commands, e-mail: docs-helphttpd.apache.org

On Jun 13, 2007, at 11:52, Joshua Slive wrote:

> It would be nice to replace the current hacked-up
google search of our
> docs with a proper "custom search engine"
from google. The problem is
> that custom search engines are tied to specific google
accounts, and I
> don't think it is a good idea for any single person to
have control
> over the config.
>
> What if we created a google account for the docs
project and stored
> its userid and password in the private pmc svn and used
that to create
> the search engine? Does this violate google's terms of
service?
> Opinions?

This sounds like a good idea.

I wonder, also, whether it would be possible to run solr,
since it's  
actually an Apache project. However, having never had any
success  
getting it working on my own servers, that might be a can of
worms.

--
If we only live,
We too will go to sea in a Sieve,---
   To the hills of the Chankly Bore!



------------------------------------------------------------
---------
To unsubscribe, e-mail: docs-unsubscribehttpd.apache.org
For additional commands, e-mail: docs-helphttpd.apache.org

> It would be nice to replace the current hacked-up
google search of our
> docs with a proper "custom search engine"
from google. The problem is
> that custom search engines are tied to specific google
accounts, and I
> don't think it is a good idea for any single person to
have control
> over the config.
>
> What if we created a google account for the docs
project and stored
> its userid and password in the private pmc svn and used
that to create
> the search engine? Does this violate google's terms of
service?
> Opinions?

Are you referring to this? http://google.com/coop/cs
e/

Looking at the examples I don't see why your suggestion
would be a
problem, though it does violate the dogfood mantra.

-- 
noodl

------------------------------------------------------------
---------
To unsubscribe, e-mail: docs-unsubscribehttpd.apache.org
For additional commands, e-mail: docs-helphttpd.apache.org

Joshua Slive wrote:
> It would be nice to replace the current hacked-up
google search of our
> docs with a proper "custom search engine"
from google. The problem is
> that custom search engines are tied to specific google
accounts, and I
> don't think it is a good idea for any single person to
have control
> over the config.
Odd, I was just thinking about this a few days ago.  But I
wanted to get 
to grips with the docs-project as a whole before putting my
size 12's 
firmly down my throat.


Does it have to be a google based SE? I am not sure if there
is any 
historical, or political reasons for this?

>
> What if we created a google account for the docs
project and stored
> its userid and password in the private pmc svn and used
that to create
> the search engine? Does this violate google's terms of
service?

Sounds like a good place to store the details.  Who would
have access to 
this? Just the PMC? People in the docs-project? or is it
more granualr 
than that?

> Opinions?
>
> Joshua.
>
>
------------------------------------------------------------
---------
> To unsubscribe, e-mail: docs-unsubscribehttpd.apache.org
> For additional commands, e-mail: docs-helphttpd.apache.org
>

------------------------------------------------------------
---------
To unsubscribe, e-mail: docs-unsubscribehttpd.apache.org
For additional commands, e-mail: docs-helphttpd.apache.org

Rich Bowen wrote:

> I wonder, also, whether it would be possible to run
solr, since it's
> actually an Apache project. However, having never had
any success
> getting it working on my own servers, that might be a
can of worms.

Firmly +1 on dogfooding.  If *we* can't get Solr to work,
who can?  I
mean it's our project, right?  Even if it's from the Java
Weenie side?

Regarding Google, who's making the money from the searches? 
I came
across a $100M figure that Firefox is making from the search
box in the
browser.  While we'd obviously not be flying that high,
would the ASF be
entitled to any such revenue from our own search box, or the
individual
who happens to own the account?

More than nit-picking over the money (I know Apache doesn't
monetize
terribly well anyway), I'm interested in the construction
an
organization like Mozilla is using vis-a-vis Google.

S.

On Wed, Jun 13, 2007 at 01:33:23PM -0400, Sander Temme
wrote:
> Rich Bowen wrote:
> 
> > I wonder, also, whether it would be possible to
run solr, since it's
> > actually an Apache project. However, having never
had any success
> > getting it working on my own servers, that might
be a can of worms.
> 
> Firmly +1 on dogfooding.  If *we* can't get Solr to
work, who can?  I
> mean it's our project, right?  Even if it's from the
Java Weenie side?
> 
Also +1 here as well. In fact I expect that something like
Solr could
work much more efficiently with our structured
documentation.

> Regarding Google, who's making the money from the
searches?  I came

I'm not too worried about the $$, but given the amount of
competition
among the search engines out there, choosing sides doesn't
seem quite
right.

vh

Mads Toftum
-- 
http://soulfood.dk

------------------------------------------------------------
---------
To unsubscribe, e-mail: docs-unsubscribehttpd.apache.org
For additional commands, e-mail: docs-helphttpd.apache.org

On 13/06/07, Joshua Slive <joshuaslive.ca> wrote:
> What if we created a google account for the docs
project and stored
> its userid and password in the private pmc svn and used
that to create
> the search engine? Does this violate google's terms of
service?
> Opinions?

Given the long history of 1.3 and the nature of google's
pagerank, I'd
fear that any search results would be weighted too much
towards the
older versions. Of course a home-grown solution could fix
that.

-- 
noodl

------------------------------------------------------------
---------
To unsubscribe, e-mail: docs-unsubscribehttpd.apache.org
For additional commands, e-mail: docs-helphttpd.apache.org

On 6/13/07, Sander Temme <sandertemme.net> wrote:
> Rich Bowen wrote:
>
> > I wonder, also, whether it would be possible to
run solr, since it's
> > actually an Apache project. However, having never
had any success
> > getting it working on my own servers, that might
be a can of worms.
>
> Firmly +1 on dogfooding.  If *we* can't get Solr to
work, who can?  I
> mean it's our project, right?  Even if it's from the
Java Weenie side?

I have no problem at all with that, providing someone is
willing to
take charge, drive it through to completion, and assure it
is
maintained for some period of time. My primary concern is
with giving
our users the best possible search. Our experience with
running our
own search engines on *.apache.org has been mixed at best.

>
> Regarding Google, who's making the money from the
searches?  I came
> across a $100M figure that Firefox is making from the
search box in the
> browser.  While we'd obviously not be flying that high,
would the ASF be
> entitled to any such revenue from our own search box,
or the individual
> who happens to own the account?

With our current search box, all the revenue is going to
google. With
CSE, since we qualify as a non-profit, we can eliminate adds
entirely.
(Or we can probably participate in their revenue sharing
thing if we
want; but again, that adds complication that I'm not looking
for at
the moment.)

With regards to who could access the account, it would
actually be all
apache comitters I believe. We could choose different access
control
if we wanted, but I don't think it is a big deal. (We just
don't want
anyone switching it to point to playboy.com... Or maybe we
do...)

With regards to 1.3 having too much weight, I would setup a
separate
search engine for each version. (I would probably also
include the
wiki in all the versions.)

Regarding google or some other public search engine, we
already use
google and they have a relatively easy system that doesn't
show adds
(and therefore earns them no direct revenue), so I don't
really see
any conflict.

(Boy, I didn't think this was going to be that complicated
an issue 

Joshua.

------------------------------------------------------------
---------
To unsubscribe, e-mail: docs-unsubscribehttpd.apache.org
For additional commands, e-mail: docs-helphttpd.apache.org

On 6/13/07, Joshua Slive <joshuaslive.ca> wrote:

> With our current search box, all the revenue is going
to google.

I take that back. Google isn't currently running adds on our
searches
it appears. But I believe they have the right to do so.

Anyway, with CSE they explicitly allow non-profits to opt
out of adds.

Joshua.

------------------------------------------------------------
---------
To unsubscribe, e-mail: docs-unsubscribehttpd.apache.org
For additional commands, e-mail: docs-helphttpd.apache.org

...

Joshua Slive wrote:
>
> I have no problem at all with that, providing someone
is willing to
> take charge, drive it through to completion, and assure
it is
> maintained for some period of time. My primary concern
is with giving
> our users the best possible search. Our experience with
running our
> own search engines on *.apache.org has been mixed at
best.

I tell you what, once a decision has been made I would be
very happy to 
take this on as my first task. (Above and beyond patching,
of course).
Is there a case for trialling so called 'rival' products in
a bid to 
ascertain their suitability for the task in hand.


...

> With regards to who could access the account, it would
actually be all
> apache comitters I believe. We could choose different
access control
> if we wanted, but I don't think it is a big deal. (We
just don't want
> anyone switching it to point to playboy.com... Or maybe
we do...)
>
> With regards to 1.3 having too much weight, I would
setup a separate
> search engine for each version. (I would probably also
include the
> wiki in all the versions.)

This should be a relatively straight forward process once we
have a 
front runner.  Enabling searching for a specific version
should be a 
rather simple approach.

>
> Regarding google or some other public search engine, we
already use
> google and they have a relatively easy system that
doesn't show adds
> (and therefore earns them no direct revenue), so I
don't really see
> any conflict.
>
>

...

--
Tony


------------------------------------------------------------
---------
To unsubscribe, e-mail: docs-unsubscribehttpd.apache.org
For additional commands, e-mail: docs-helphttpd.apache.org