The DB layer won't help, certainly. But I consider the trade-off in performance for that component on its own quite acceptable for the actual increase in security it brings (considering third party modifications etc.)
I'm sure everyone will agree
:)The other major performance headache is post moderation. There's no way around it, it is a performance headache if it is turned on.
And this is what still makes me unsure how to implement post privacy, i.e. what is basically stalling Wedge the most and caused me to post a call for help in the shape of a haiku...
:lol: (Because I'm preeeeetty sure that once we get this done, Wedge is 99% ready for a public alpha...)
For all other situations, you'd need to compare them with as-close-to-identical setup, so any optimisation work you did for 1.x (e.g. turn off calendar, DB tables being InnoDB and appropriate table structure changes to convert from tinytext to varchar) would need to be mirrored in 2.x.
I'm still sure it's mostly about perceived performance. It's best to test with PageSpeed or YSlow, and of course also enable benchmarking in SMF and check the query logs to see if generation times on the server are pretty much the same or not. If they're similar, then it's a problem with client-side caching, multiple JS/CSS files and so on. (i.e. my domain of choice 8))
Also if you haven't already, switch the avatars to their own folder.
Here's wondering if we're already doing it by default in Wedge... If not, we probably should
:PPosted: May 14th, 2012, 12:23 PM
And this is what still makes me unsure how to implement post privacy, i.e. what is basically stalling Wedge the most and caused me to post a call for help in the shape of a haiku... :lol: (Because I'm preeeeetty sure that once we get this done, Wedge is 99% ready for a public alpha...)
Forgot about custom groups... That one is annoying, too: I still don't know if it's realistic to keep having a list of membergroups in the member table, or if we should always join another table.
(Crap... If only I had the balls to say "let's postpone this to a later version" and actually STICK to it and make a converter later...)