GDBus benchmarks or it didn't happen (Was Re: "DBus Embedded" - a clean break)

[David Zeuthen]

Hi,

On Thu, Jan 27, 2011 at 4:59 AM, Michael Meeks <michael.meeks at novell.com> wrote:
>
> On Wed, 2011-01-26 at 09:58 -0500, David Zeuthen wrote:
>> I'll repeat what I've said earlier: if you want low latency / high
>> bandwidth, the D-Bus protocol greatly limits what you can do
>
>        This is clearly both true, and not that helpful :-) People don't need
> something 10,000 times faster, only ~ten - and that is quite achievable
> by optimising GDBus IMHO :-)

So if D-Bus is 1% of your resource usage and you make the
implementation 10 times faster you went from 1% to 0.1%. Sorry, but
that's almost a waste of time. OTOH, if D-Bus is 30% of your resource
usage, changes are you need changes on a higher level (redesign your
app / high-level protocol). Maybe you don't want to do these changes.

Btw, who _are_ these people that want GDBus to be faster? There's
exactly one bug [1] filed about GDBus performance (see elsewhere in
the thread for the bug reference) and the only benchmark it has is not
very good. As you can tell, I actually did quite a lot of work to try
and work with the reporter and I even pointed out where things could
be improved.

And, gee, it's not like I hate performance. It's just that there's no
_hard_ data explaining what the problem is with GDBus or even if there
is a problem. There is, however, a lot of 'gut feelings' about GDBus
and a lot of useless theories and a lot of "let's try to optimize
this" etc. Show me the data. Then we'll work on fixing the problems
with (if there are any).

And please do continue this discussion in the GNOME bugzilla - it's
useless to keep ranting here.

     David

[1] : https://bugzilla.gnome.org/show_bug.cgi?id=634471