Hi all

Now that Y! has thrown us to the wolves, (requireing quickfix form
salathe to get it running again), its probably good time to look into
alternative searching for phpweb.


Currently phpweb has several search possibilities:
  - function list
42y old sqlite search

  - general mailing list
  - developer mailing list
  - documentation mailing list
mailinglist (archives) search - which IMO is totally different
subject, and really irrelevant for that particular searchbox on
phpweb.

  - bug database
redirected to bugsweb (doesn't even actually trigger a search, just
prefills the form) - no idea why thats an option in that searchbox

  - online documentation
  - all php.net sites
  - this mirror only
  - Site News Archive
  - All Changelogs
  - just pear.php.net
  - just pecl.php.net
  - just talks.php.net
Y!/Bing/Wtf search.


I'd like to kill of the external search.. people can simply use their
preferred search engines when searching the websites.

As for the function list (aka php.net/keyword), and online doc
search.., I'd like to use the PhD database.
PhD already stores the refname and refpurpose, and has all the legacy
capabilities (by filename/id) too..
The only disadvantage of using the PhD database rather then the
magically generated database (that only handful of people actually
know where and how is generated), is the fact it uses SQLite3.. and
mirrors randomly have ext/sqlite, ext/sqlite3, ext/pdo_sqlite,
ext/pdo_sqlite3, ext/wtf_sqlite, ext/other_sqlite3,
ext/and_maybe_other_sqlite_extension,
hext/ow_may_sqlite_exts_are_there_really... which means we would have
to create some sort of abstraction layer - and maybe even duplicate
the sqlite3 as sqlite database..
Also, the generated database itself doesn't actually support i18n, but
thats easily workedaround because we know which language we are
generating the manual for (on the rsync box) each time, so we can
copy/rename the db file accordingly.

What do you guys think?
Should we killoff random searches (and the option from my.php) and use
the PhD db?

We could also improve(add) indexterm support to the PhD indexer, and
add indexterms to various doc pages to improve search capabilities..
giving us finegrained control over search results from withing the
docs themselfs..

-Hannes

Reply via email to