<?xml version="1.0" encoding="windows-1252"?>
<node id="1014038" title="Re^10: Writing a database lookup tool" created="2013-01-18 06:45:26" updated="2013-01-18 06:45:26">
<type id="11">
note</type>
<author id="846903">
elef</author>
<data>
<field name="doctext">
Thanks, that is interesting.&lt;br&gt;
Your query times give me pause. Ideally, I'd be shooting for about 1sec on most searches. If 5+ second query times are the norm, that kind of kills it because this thing would be used for quite frequent lookups. Obviously, I don't want to spend weeks building this thing only to find out that it's too slow to be practical when I load it up with all my data, so I'll have test any tool I consider with all the data before I go ahead.
&lt;br&gt;Now, many searches could be done as full word searches. Would those be radically faster than general substring searches if I were to use sql via DBI:DB or something similar, or would I need to use something that was optimised for text search (e.g. Lucene/Solr) to get that performance benefit? I posted a question on FTS in reply to NetWallah's post further down that's kind of related.
&lt;br&gt;&lt;br&gt;
At this point the whole thing is just an idea I'm considering. I guess I will play with some of the tools suggested here, and maybe look into completely different approaches like trying to use libreoffice base.&lt;br&gt;</field>
<field name="root_node">
1011632</field>
<field name="parent_node">
1011765</field>
</data>
</node>
