[PATCH] Remove obscure "low-latency" parts in the introduction of spec

[Simon McVittie]

On 29/05/13 21:24, Justin Lee wrote:
> Therefore, latency is unrelated to whether the operation is
> asynchronous or synchronous.

Indeed. This patch seems fine to me; any opinions from other maintainers?

I think when the author of that section of the spec talks about avoiding
round-trips, what they mostly mean is avoiding *blocking on*
round-trips. Messages having a reply is fine, but it becomes a problem
if you can only do one thing at a time.

This is how D-Bus is meant to be used:

    start call to DoSomeThing
    go back to the main loop
    carry on, do other useful work, re-draw the GUI, whatever
        (perhaps including starting more D-Bus calls)
    ... time passes ...
    receive reply from DoSomeThing; if it failed, recover
        (pop up an error or retry or whatever is appropriate)

and this is a bad pattern that hurts performance:

    start call to DoSomeThing
    wait for reply from DoSomeThing; if it failed, recover
        (pop up an error or retry or whatever is appropriate)
    carry on with the rest of your application