I knocked off two items on my TODO list today – its amazing how many small tasks build up when you actually maintain a list fully…:-(

First up, I complete the PHP DataObjects tutorial, which is now languishing on the Devnetwork Tutorial Proposals forums awaiting final comments. It comes to about 6,300 words (excluding code) so its pretty lengthy (wordy more like it) but it looks good, reads okay and serves its purpose. Someone owes me a lot of free pints for this…

Secondly, I dug into reviewing Scuttle, an open source project to create a links archive along the lines of del.icio.us. The SF page is over at http://sourceforge.net/projects/scuttle/. I found a few potential problems in the code where the DB layer lacks more in-depth accurate escaping methods. Duly reported, and hopefully fixed in the next version. I may install to scuttle.quantum-star.com as a more focused personal links archive for community members. More code to read through, so maybe more bugs to report later.

Thirdly, I drew up a few simple designs for extending the Partholan DataAccess class. Basically I’m hoping to add multi-row update support and WHERE condition ranges to the mix. These would solve the vast majority of needs in QSE in areas needing them without sacrificing too much to complexity in the API. They’ll also allow further performance gains in the system as a whole – not that it’s a huge worry.

I long ago made the call we needed simplicity over highly specific SQL which would perform a lot better. We are making gains in other areas which offer far more payback. Afterall an alternative to speeding up SQL queries is to limit the number of such queries required in the first place. The new Ship Model is a classic example. In Solar Empire thee is a lot of cross pollination of data across tables, simply to track locations and such. In QSE, we simplify such tracking down to IDs. For example, in QSE location is determined by fleet, which in turn determines the location of any ship based on a fleet_id field which rarely changes. Solar Empire on the other hand has no aggregate grouping mechanism, and must therefore track locations for every individual ship in the game.

The mix of DataAccess and an improved efficient database design is plain better than sticking with the old ways. It definitely performs better compared to QS2′s system.

Related posts:

  1. A QS Evolved progress check…
  2. QS Evolved: The Ship Model
  3. ActiveRecord for Partholan
  4. Progress Report (Better late than never!)