Berkeley DB format

> I don't mean to be rude, but folks, the decision for 1.0 has been made -
> we're using Berkeley DB. If you want to keep postulating about
> whether a text-based database could be used, feel free, but please do it
> elsewhere. Come back here when you've got running code, and we can
> consider it for later. Brian
   
   (replying off a post read on the web interface, I hope the web
   archive isn't lagged and this discussed to death)

   I think using a Berkeley backend is just fine.. it's well tested, and easy to
   load into any scripting language to stick tools in front of (Ruby,
   Perl, PHP, C, etc. come to mind).

   I think, due to the performance hit, and rare-times where text is
   actually needed, the best course is to just use BDB as the backend,
   and have an export/import script to spit it into a plain text format
   to do manipulations, in the extremely rare time that it's needed.

   If we're going to be guaranteeing things with MD5, and using BDB, it
   should be very rare that you'll have corruption and have to get in
   there, after all. Using an import/export script (assuming it can
   deal with the corruption), should not be that difficult.

   Just my 2 kronor.

-- 
: Thomas Stromberg                      work> tstromberg@rtci.com  :
: Research Triangle Commerce (ICC.net)  home> thomas@stromberg.org :
'Every word is like an unnecessary stain on silence and nothingness' 
    -- Beckett