Return-Path: Received: from vid.zoo.gr ([206.188.4.140] verified) by mx.softsearch.ru (CommuniGate Pro SMTP 5.1.14) with ESMTP id 64245544 for talk@openhack.ru; Fri, 22 Feb 2008 18:09:55 +0300 Received: from [192.168.1.100] (ip82-139-112-245.lijbrandt.net [82.139.112.245]) (Authenticated sender: kostas) by vid.zoo.gr (Postfix) with ESMTP id E8BEA2180E2 for ; Fri, 22 Feb 2008 17:09:55 +0200 (EET) Message-ID: <47BEE5C7.7090709@chatzi.org> Date: Fri, 22 Feb 2008 16:09:59 +0100 From: Kostas Chatzikokolakis User-Agent: Thunderbird 2.0.0.6 (X11/20071022) MIME-Version: 1.0 To: talk@openhack.ru Subject: Re: Reconnections and mod_perl References: In-Reply-To: X-Enigmail-Version: 0.95.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit > Using explicit namespace => 'prefix' is different from adding 'prefix' > by hand: namespace string is not hashed along with the key, thus keys > are distributed among servers differently. But if all parts of your > application agree to use the same approach, then go ahead, hashing > (C::M::F uses CRC32) is fast enough, so additional 16-32 bytes won't > make a noticeable slowdown. Good to know. I think it should be more clear in documentation (does Cache::Memcached work also this way?) cause someone could think he can use both interchangeably. Another problem with manually adding prefixes is that you also have to remove them in the result of *_multi. A good solution would be to change the value of namespace before each call to get/set. I guess this is not currently possible, right? It would be a good idea to add it in a future version (so that you can reuse the same object (and its connections) for different namespaces). Kostas