[wp-hackers] Term Meta - Trac'd already?

[Mike Schinkel]

On Jul 13, 2011, at 11:42 PM, Otto wrote:
> Meh. I couldn't say for sure what "best practice" is in this case. I know
> how I'd do it. It'd work. It'd scale. It'd get the job done for at least 5
> years, which is all my timeline really is anymore. Modern code doesn't need
> to last longer than that. We're not writing in C and expecting things to
> remain constant for 30 years here.

To be clear, I wasn't referring to best practices regarding longevity for 5 or 30 years, I was referring to what makes it easy to build, launch and maintain a site today, and for the foreseeable future of the site, typically 1 to 36 months, 72 at the high end.

> Depends on how you define "complex". My queries would be simpler mainly
> because I wouldn't write a single query. I'd use existing core functions
> throughout.

More complex in that you'd require more indirections in the PHP code to get there vs. taxonomy meta.

> If you want to look at it from that angle, then roughly 48% of all WordPress
> sites scale like that. Don't forget wp.com and it's 26 million sites. 7% of
> the web is a big chunk of change. :)

Apples and Oranges. WordPress.com is sharded; millions of individuals tables, not a few tables with millions of tenants.  
So WordPress.com != Facebook/Twitter/Google in architecture.

OTOH I do completely understand from WordPress.com why Automattic doesn't want to have to change the table structures for millions of sites. That's a valid, tangible, no bike-shedding concern.  I get that. :)

-Mike