[prev in list] [next in list] [prev in thread] [next in thread] 

List:       puppet-users
Subject:    Re: [Puppet-users] File bucketing changed?
From:       Luke Kanies <luke () madstop ! com>
Date:       2007-03-30 19:39:59
Message-ID: DEFAFC73-315D-4B59-9ECE-FEF193299C5E () madstop ! com
[Download RAW message or body]

On Mar 30, 2007, at 9:35 AM, Benjamin C. Kite wrote:

>
> On Mar 24, 2007, at 5:28 PM, Luke Kanies wrote:
>
>> Sorry; should have released both at the same time.
>>
>> There is now a 'filebucket' script available in svn; it'll be up to
>> the packagers to get it into the packages.
>>
>> Unfortunately, in the process of writing this, I realized I made a
>> stupid mistake in setting up nested paths in the filebucket:  I was
>> only using the first 8 chars of the md5 sum, not the whole thing
>> (e.g., a file would be in a/b/c/d/e/f/g/h/contents, rather than /a/b/
>> c/d/e/f/g/<md5>/contents).
>>
>> That's now fixed, but until I release that (probably in the next
>> couple of days), there's a chance of conflicts in the filebuckets.
>
> I'm curious why (in particular) this was viewed as a mistake.
>
> Obviously there's efficiency to be gained from sizing your hash tree
> properly...

Sorry; my examples weren't the same length, but that wasn't really  
the problem:  The problem was the first example did not include full  
md5 sum, and the second did.

Collision likelihood for the full md5 is darn-near impossible, but if  
you only include the first 8 characters of the sum, suddenly it  
becomes a lot more likely.

  --
  Talent hits a target no one else can hit; Genius hits a target no one
  else can see.  -- Arthur Schopenhauer
  ---------------------------------------------------------------------
  Luke Kanies | http://reductivelabs.com | http://madstop.com



[prev in list] [next in list] [prev in thread] [next in thread]