Forum .LRN Q&A: Re: Bug in search with postgres

Collapse
2: Re: Bug in search with postgres (response to 1)
Posted by Don Baccus on 07/20/07 11:28 AM
OpenACS 5.3.0's search initialization script always scheduled the Oracle indexer regardless of database, which was a bug.

OpenACS 5.3.2 (which has just been built) fixes that bug.

If you don't want to update, just go to

search/tcl/search-init.tcl

and delete the call which schedules the tcl proc which calls the Oracle Ctx* proc, and restart your server.

Collapse
3: Re: Res: Re: Bug in search with postgres (response to 2)
Posted by Jose Agustin Lopez Bueno on 07/20/07 12:04 PM
Ok,

but I am seeing the code and I think of that that code is running, both Postgres and Oracle:

ad_proc -private search::indexer {} {
Search indexer loops over the existing entries in the search_observer_queue
table and calls the appropriate driver functions to index, update, or
delete the entry.

@author Neophytos Demetriou
@author Jeff Davis (mailto:davis@xarg.net)
} {

set driver [ad_parameter -package_id [apm_package_id_from_key search] FtsEngineDriver]

if { $driver eq ""
|| (![callback::impl_exists -callback search::index -impl $driver] && ! [acs_sc_binding_exists_p FtsEngineDriver $driver])} {
# Nothing to do if no driver
ns_log Debug "search::indexer: driver=$driver binding exists? [acs_sc_binding_exists_p FtsEngineDriver $driver]"
return
}
# JCD: pull out the rows all at once so we release the handle
foreach row [db_list_of_lists search_observer_queue_entry {}] {
nsv_incr search_static_variables item_counter
if {[nsv_get search_static_variables item_counter] > 1000} {
nsv_set search_static_variables item_counter 0
db_exec_plsql optimize_intermedia_index {begin
ctx_ddl.sync_index ('swi_index');
end;
}
}
...