Flat-file systems vs Software DBMS

[Gigaman2003]

So since this forum has been rather quiet lately, I thought i'll post this to generate some discussion.

The news:
PQL, TDB and txtSQL are doing a joint benchmark! The aim of this benchmark is to give ourselves a better look at how we (flat-file DBMSes) compare to each other and mysql. Not going to delay this anymore, just check out what im talking about. This is incomplete and only contains the PQL benchmark. The txtdb and txtsql equivalents will be written by the authors of those systems. txtdb is at version 5.2, txtsql at 2.2 and pql, being the youngest, is at 0.7.Anyway it'll give you an idea of where i've gotten with PQL since version 0.4.

Some of you might recall I've brought up PQL here before (thread probably already archived), but since then , its improved by leaps and bounds. The querying capability of PQL still is my proudest acheivement, and I've made several very cool innovations to increase the speed of the queries 20 times. There is also a (almost global) shadow cache system in place to control memory leaks and unnecessary file reads that killed 0.4's speed.

I can't say anything about txtdb and txtsql cause i don't know them as well, but we are all definately very interested to see the final results of this benchmark. It should change the way people think about flat-file systems (buggy, feature poor, slow, prone to corruption). What do you think?