[ANN] FastRI 0.2.1: faster full-text searching, multiple queries, pager, class/method lists

FastRI is an alternative to the ri command-line tool. It is *much* faster, and
also allows you to offer RI lookup services over DRb. FastRI is smarter than
ri, and can find classes anywhere in the hierarchy without specifying the
"full path". FastRI can perform full-text searching. Its RubyGems support is
better than ri's, and it knows which gem a method/class definition came from.

Getting it

···

==========
Additional information, tarballs... at
  http://eigenclass.org/hiki.rb?fastri

FastRI can be installed with RubyGems:
  gem install fastri
(if you get an old version/a 404 error, please allow some time after the
release until the package propagates to the RubyForge mirrors). Please read
below for an important note regarding the RubyGems packages.

User-visible changes since version 0.2.0

* fri will use a pager in full-text search mode, and also when asked to
  (-P, -T, --[no-]pager, --pager-cmd)
* multiple queries supported: fri upcase downcase
* fri -S is 100ms faster since it avoids requiring rubygems.rb
* list all classes/modules and methods with --classes and --methods, and both
  with --list-names (-l)

Usage

There are two parts to FastRI:
* the server: fastri-server
* the client: fri

FastRI uses a Rinda Ring to allow servers to be discovered automatically
without needing to indicate the DRb URIs manually. It can work across
machines if you make sure the ring server is bound to the correct interface,
and the ACL permissions are correct.

Examples

$ fastri-server (creates the index on the first run, blocks)

Later, (times measured with a cold cache):
$ time ruby bin/fri -f plain Array#fetch
------------------------------------------------------------ Array#fetch
      array.fetch(index) -> obj
[...]
real 0m0.287s (real 0m0.127s with a hot cache)
user 0m0.048s
sys 0m0.008s

Compare to:
$ time ri -T -f plain Array#fetch
------------------------------------------------------------ Array#fetch
[...]
real 0m10.136s (real ~ 1.5s with a hot cache)
user 0m1.140s
sys 0m0.464s

This illustrates FastRI's ability to locate classes deep in the class
hierarchy:

$ fri Base
------------------------------------------------------ Multiple choices:

      ActionMailer::Base, ActionView::Base, ActionWebService::API::Base,
      ActionWebService::Base, ActionWebService::Client::Base,
      ActiveRecord::Base, MapReduce::ActiveRecord::Base,
      RSS::Maker::Base, Scruffy::Components::Base,
      Scruffy::Formatters::Base, Scruffy::Layers::Base,
      Scruffy::Renderers::Base, Scruffy::Themes::Base

$ fri Themes::Base
------------------------------------------- Class: Scruffy::Themes::Base
      Scruffy::Themes::Base
      Author: Brasten Sager

      Date: August 14th, 2006

Compare to
$ ri Themes::Base .... several seconds later ...
Nothing known about Themes::Base

A small note about RubyGems + FastRI.

RubyGems adds a noticeable overhead to fri, making it run slower than if you
installed it directly from the tarball with setup.rb.

Compare the execution time when installed with RubyGems:
  $ time fri -f plain String > /dev/null

  real 0m0.385s
  user 0m0.244s
  sys 0m0.036s

to the time fri actually takes to run, without the overhead introduced by
RubyGems:
  $ time ruby bin/fri -f plain String > /dev/null

  real 0m0.088s
  user 0m0.040s
  sys 0m0.008s

If you care about those extra 300ms (and there are situations where they will
matter, e.g. when using fri for method completion), get FastRI from the
tarballs.

License

FastRI is licensed under the same terms as Ruby. See LICENSE.

Feedback

Bug reports, patches, comments... are appreciated.
You can contact the author via <mfp@acm.org>. Please add "fastri" to the
subject in order to bypass the spam filters.

--
Mauricio Fernandez - http://eigenclass.org - singular Ruby

Thanks, this looks nice. However,

$ fri flatten
------------------------------------------------------ Multiple choices:

     Array#flatten, Set#flatten
$ fri A#flatten
nil

so I need to fully spell out which one I'm interested in. Could you
please add a simple string distance or something like that so that one
can use shortcuts, ie fri A#flatten should choose Array#flatten since
it is the closest one. If I don't misremember we discussed and added
this functionality to ri way back then.

/Robert Feldt

···

On 11/23/06, Mauricio Fernandez <mfp@acm.org> wrote:

FastRI is an alternative to the ri command-line tool. It is *much* faster, and
also allows you to offer RI lookup services over DRb. FastRI is smarter than
ri, and can find classes anywhere in the hierarchy without specifying the
"full path". FastRI can perform full-text searching. Its RubyGems support is
better than ri's, and it knows which gem a method/class definition came from.

>FastRI is an alternative to the ri command-line tool. It is *much* faster,
>and also allows you to offer RI lookup services over DRb. FastRI is smarter
>than ri, and can find classes anywhere in the hierarchy without specifying
>the "full path". FastRI can perform full-text searching. Its RubyGems
>support is better than ri's, and it knows which gem a method/class
>definition came from.
>
Thanks, this looks nice. However,

$ fri flatten
------------------------------------------------------ Multiple choices:

    Array#flatten, Set#flatten
$ fri A#flatten
nil

so I need to fully spell out which one I'm interested in. Could you
please add a simple string distance or something like that so that one
can use shortcuts, ie fri A#flatten should choose Array#flatten since
it is the closest one. If I don't misremember we discussed and added
this functionality to ri way back then.

I have pushed a couple patches to HEAD
(http://eigenclass.org/repos/fastri/head\) that implement these search methods:
* complete namespace (m)
* complete both namespace and method (f)

FastRI's search strategy can be specified with the -O option:

    -O, --order ORDER Specify lookup order.
                                     (default: eEnNpPxX)
                                     Uppercase: case-indep.
                                     e:exact n:nested p:partial (completion)
                                     x:nested and partial m:complete namespace
                                     f:complete both class and method

See [225037] for an explanation of the basic search modes.

You can make fri behave like ri (regarding partial completion of the namespace
name) by adding 'm' to the lookup order:

$ fri -Om A#flatten
---------------------------------------------------------- Array#flatten
     array.flatten -> an_array

Moreover,

$ fri -Of A#a
------------------------------------------------------ Multiple choices:

     ACL#allow_addr?, ACL#allow_socket?, Abbrev#abbrev, Array#abbrev,
     Array#assoc, Array#at, Autotest#add_sigint_handler,
     Autotest#all_good

However, I'm not sure about the position in the search strategy (currently
eEnNpPxX) at which the 'm' or 'f' modes could be added.
Maybe eEnNpPxXmM(fF)a ?

Anyway, even if I left the default search strategy unchanged, it could be
overridden with --order (-O).

···

On Fri, Nov 24, 2006 at 12:46:36AM +0900, Robert Feldt wrote:

On 11/23/06, Mauricio Fernandez <mfp@acm.org> wrote:

                                     a:match method name anywhere

--
Mauricio Fernandez - http://eigenclass.org - singular Ruby

I have pushed a couple patches to HEAD
(http://eigenclass.org/repos/fastri/head\) that implement these search methods:
* complete namespace (m)
* complete both namespace and method (f)

FastRI's search strategy can be specified with the -O option:

    -O, --order ORDER Specify lookup order.
                                     (default: eEnNpPxX)
                                     Uppercase: case-indep.
                                     e:exact n:nested p:partial (completion)
                                     x:nested and partial m:complete namespace
                                     f:complete both class and method
                                     a:match method name anywhere

See [225037] for an explanation of the basic search modes.

You can make fri behave like ri (regarding partial completion of the namespace
name) by adding 'm' to the lookup order:

$ fri -Om A#flatten
---------------------------------------------------------- Array#flatten
     array.flatten -> an_array

Moreover,

$ fri -Of A#a
------------------------------------------------------ Multiple choices:

     ACL#allow_addr?, ACL#allow_socket?, Abbrev#abbrev, Array#abbrev,
     Array#assoc, Array#at, Autotest#add_sigint_handler,
     Autotest#all_good

However, I'm not sure about the position in the search strategy (currently
eEnNpPxX) at which the 'm' or 'f' modes could be added.
Maybe eEnNpPxXmM(fF)a ?

Anyway, even if I left the default search strategy unchanged, it could be
overridden with --order (-O).

Ok, great. If it could check some ENV variable for the search order
people could spec their own default. FASTRI_LOOKUP_ORDER? I don't
think that would add very much to the execution time...

Regards,

Robert

Wouldn't a simple shell alias do?

···

On Nov 23, 2006, at 6:03 PM, Robert Feldt wrote:

Ok, great. If it could check some ENV variable for the search order
people could spec their own default. FASTRI_LOOKUP_ORDER? I don't
think that would add very much to the execution time...

Yeah, that's right. Too late in the day over here... :wink:

Regards,

/RF

···

On 11/23/06, Xavier Noria <fxn@hashref.com> wrote:

On Nov 23, 2006, at 6:03 PM, Robert Feldt wrote:

> Ok, great. If it could check some ENV variable for the search order
> people could spec their own default. FASTRI_LOOKUP_ORDER? I don't
> think that would add very much to the execution time...

Wouldn't a simple shell alias do?