From: Jim Cain <jecain@...>
Date: 2007-04-03T00:44:30+09:00
Subject: Re: Using transactions with secondary indexes in bdb

OK, I figured out that a lot of my previous code was wrong, not even
considering the transactions, so I removed the transactional stuff and
got down to this:

require 'bdb'

BDB::Hash.open('models.db', nil, BDB::CREATE | BDB::TRUNCATE) {}
BDB::Hash.open('makes.db', nil, BDB::CREATE | BDB::TRUNCATE,
:set_flags => BDB::DUP | BDB::DUPSORT) {}

BDB::Hash.open('models.db', nil, BDB::CREATE) do |models|
  BDB::Hash.open('makes.db', nil, BDB::CREATE, :set_flags => BDB::DUP
| BDB::DUPSORT) do |makes|
    models.associate(makes) { |sdb, key, value| value }

    puts '===== Before'
    puts '----- models'
    models.each { |key, value| puts "#{key} = #{value}" }
    puts '----- makes'
    makes.each_primary { |skey, pkey, pvalue| p [skey, pkey, pvalue] }

    models['Civic'] = 'Honda'
    models['Accord'] = 'Honda'
    models['S2000'] = 'Honda'
    models['Camry'] = 'Toyota'
    models['Solara'] = 'Toyota'

    puts '===== After'
    puts '----- models'
    models.each { |key, value| puts "#{key} = #{value}" }
    puts '----- makes'
    makes.each_primary { |skey, pkey, pvalue| p [skey, pkey, pvalue] }
  end
end

And here is the output, which looks right to me:

===== Before
----- models
----- makes
===== After
----- models
Civic = Honda
Accord = Honda
Camry = Toyota
Solara = Toyota
S2000 = Honda
----- makes
["Honda", "Accord", "Honda"]
["Honda", "Civic", "Honda"]
["Honda", "S2000", "Honda"]
["Toyota", "Camry", "Toyota"]
["Toyota", "Solara", "Toyota"]

Now I just have to figure out the transactional part. I'll follow up
on the list once I do.

Cheers,
Jim