jmettraux / rufus-tokyo Goto Github PK

http://groups.google.com/group/rufus-ruby/browse_thread/thread/a96eed0d87bbd5f2

.

as requested by Ilya...
(double check with Edo, it might be a Tokyo-only issue)

I tried this on 2 machines, with a variety of versions of ffi and tokyo cabinet. It works in 1.8, but does not in 1.9, failing on the line 'db.close':

require 'rubygems'
require 'rufus/tokyo'

def test_tokyo
  db = Rufus::Tokyo::Cabinet.new('/tmp/data.tch', :type => :hash,
                                                  :mode => 'wc', # create/write
                                                  :opts => 'ld')
  db['nada'] = 'surf'
  puts "Should be 'surf': #{db['nada']}"
  puts "Should be 'nil': #{db['xyz']}"
  5000.times { |i| db[(i+1).to_s] = "x" }
  puts "Should be '5000': #{db.inject { |r, (k, v)| k }}"
  closed = db.close
  puts "Should be 'true': #{closed}"
  raise "Could not close db" if !closed # this line really pukes in 1.9
end

5.times {test_tokyo; sleep 0.5}

http://groups.google.com/group/rufus-ruby/t/2ad37d7864c55155

don't fail or at least educate lib user...

http://groups.google.com/group/rufus-ruby/t/a8f057740e645aeb

find a way to list indexes set on a table.

http://groups.google.com/group/rufus-ruby/t/a8f057740e645aeb

1.8.6 FFI 0.5.0

as discussed with Zev

rescue block bypassed, probably error class changed

as per chat with swistak

thanks Kenneth Kalmer

.

http://groups.google.com/group/rufus-ruby/t/b0ee0d76fd01e95c

I tried to run example with the creation of a new table and here is what I got:

require 'rubygems'
require 'rufus/tokyo'
t = Rufus::Tokyo::Table.new('table.tct')
Rufus::Tokyo::TokyoError: (err 0) success
from /home/lumrandir/.rvm/gems/ruby/1.8.6/gems/rufus-tokyo-1.0.1/lib/rufus/tokyo/cabinet/table.rb:603:in raise_error' from /home/lumrandir/.rvm/gems/ruby/1.8.6/gems/rufus-tokyo-1.0.1/lib/rufus/tokyo/cabinet/table.rb:588:inlibcall'
from /home/lumrandir/.rvm/gems/ruby/1.8.6/gems/rufus-tokyo-1.0.1/lib/rufus/tokyo/cabinet/table.rb:136:in initialize' from (irb):14:innew'
from (irb):14
from /home/lumrandir/.rvm/ruby-1.8.6-p383/lib/ruby/site_ruby/1.8/rubygems/exceptions.rb:82

I ran it on Ruby 1.8.6-p383, but on Ruby 1.9.1 and Ruby Enterprise Edition 1.8.7 I have same error.

Environment:
rufus-tokyo 1.0.1
ffi 0.5.1
tokyocabinet 1.4.36
Slackware 13.0 x86

The current GEM attempts to install ffi on jruby thus causing the install to fail.

http://groups.google.com/group/rufus-ruby/browse_frm/thread/a8f057740e645aeb

faster deletes

Tyrant supports the hot backup feature and disabling it inside of rufus-tokyo doesn't provide any security. The feature is still active in the server and could be reached through a simple HTTP request. Forcing me to resort to that if I need the feature is just an inconvenience.

Sure, this feature can be dangerous. If you need to restrict access to it though, you want to do that at the server level so it affects everyone equally. This is the purpose of the -mask and -umask switches ttserver supports. You could also setup access to Tokyo Tyrant through something like SSH tunneling to restrict it to specific trusted users.

Thus, I believe disabling copy() is costing a feature without a noticeable gain.

Hi, I've got the next error when I was adding data to the DB. I'm using the last version downloaded and compiled fom
http://sourceforge.net/projects/tokyocabinet/files/

/var/lib/gems/1.9.0/gems/rufus-tokyo-0.1.14/lib/rufus/tokyo.rb:48:in `blen': undefined method `size' for nil:NilClass (NoMethodError)
    from /var/lib/gems/1.9.0/gems/rufus-tokyo-0.1.14/lib/rufus/tokyo/cabinet/util.rb:98:in `[]='
    from /var/lib/gems/1.9.0/gems/rufus-tokyo-0.1.14/lib/rufus/tokyo/cabinet/util.rb:191:in `block in from_h'
    from /var/lib/gems/1.9.0/gems/rufus-tokyo-0.1.14/lib/rufus/tokyo/cabinet/util.rb:191:in `each'
    from /var/lib/gems/1.9.0/gems/rufus-tokyo-0.1.14/lib/rufus/tokyo/cabinet/util.rb:191:in `inject'
    from /var/lib/gems/1.9.0/gems/rufus-tokyo-0.1.14/lib/rufus/tokyo/cabinet/util.rb:191:in `from_h'
    from /var/lib/gems/1.9.0/gems/rufus-tokyo-0.1.14/lib/rufus/tokyo/cabinet/util.rb:203:in `[]'
    from /var/lib/gems/1.9.0/gems/rufus-tokyo-0.1.14/lib/rufus/tokyo/cabinet/table.rb:227:in `[]='
    from main.rb:26:in `block in <main>'
    from main.rb:24:in `each'
    from main.rb:24:in `<main>'

as seen in http://twitter.com/neilfws/status/6889890828

"I was opening the database and then doing an exec, which resulted in the table being locked until the exec'd process ending."

via #misc

after a certain timeout, check connection before next op.
instrument libcall / wrap libcall ?

What about EDO ? (don't forget tables)

results in a full table scan apparently

http://groups.google.com/group/rufus-ruby/t/a8f057740e645aeb

Got a segmentation fault on TokyoCabinet B+ when doing Tokyo::Cabinet#getdup at non-existant key.

Way to reproduce:
db = Rufus::Tokyo::Cabinet.new('data.tcb')
db.putdup('a', 'v')
p db.getdup('a') # fine
p db.getdup('b') # segfault
db.close
Thanks!

http://sourceforge.net/project/shownotes.php?group_id=200242&release_id=691399

I realize it's not a proper query option yet, which is why I hesitated to add it myself in the changes I made. Still, it seems like a nice convenience from user code. What do you think?

http://tokyocabinet.sourceforge.net/tyrantdoc/#tcrdbapi

following http://github.com/nofxx/rufus-tokyo/commit/d07d480b06c0708353cd1e99ba8dd6e605d8ae28

or the reverse

http://groups.google.com/group/rufus-ruby/t/5b134b975d2697cc

• EDO
• tyrant : no support