[wp-hackers] Posting still slow
scott at scottberkun.com
Wed Aug 31 21:22:02 GMT 2005
> Matt Mullenweg wrote:
> I'm less inclined to have some sort of progress saying "You have pinged
> X... Y... Z..." because (1) it would be a lot more complicated to
> implement and more importantly (2) no one really cares. It's something
> that should just happen without being in your face or interrupting your
Agreed. Best case is fixing actual perf so this is a non issue.
But if that's not possible (which given random net latency seems the case),
there are subtle non-annoying ways to make latency seem shorter and make
clear that everything is still ok. If I knew more about the details I could
mock some up.
Do we have any actual perf data? Or a breakdown on where the perf sinks
are within the post process?
Sample size of one: I'd say one post in ten in my current install has a
several second lag. I have no idea why: could be WP, DSL, my ISP, or my dog
Max redlining bandwith by downloading porn. But when it happens all I see is
the WP screen - so guess who gets blamed first :)
Just fyi: There's evidence that percieved latency works at 3 levels:
1/10th second, 1 second, and 10 seconds. If we had data on average latency
times for different situations, we'd know where we're at and how much we
even need to care. It's when you're in the 1.1 to 10 second range that some
kind of visual feedback makes a difference.
 I've worked on projects with a no-frills perf lab: a dedicated machine
than ran test cases on a special debug build, pumping out raw time data used
to answer perf questions for different versions/installs/etc. This minimized
subjective arugments about what feels slow or not and helped measure perf
improvements. I don't expect we have that here, but it seemed worth
More information about the wp-hackers