Hello all,
Timings per link would be absolutely fantastic, but due to the underlying
architecture of FME they are very hard to get.
I will spend more time on this thought experiment.
<time elapses>
Okay, I think it would be more useful to label each transformer itself with the
time that was spent inside of it. In fact, I don't think one could label
the links with the time, as several links may go in OR out of a particular
transformer, and the transformer is really where the time is spent.
But I still can't think of a good way to extract the CPU time spent in a
particular transformer or FME factory.
<more time elapses>
I'm giving up for tonight.
Dale
On 1/24/06, mark2atsafe
<[EMAIL PROTECTED]> wrote:
Jason,
I think timings would be a lot more difficult. Depending on the
combination of transformers a single feature usually passes through
the whole workspace at once, rather than all features through a
single transformer at a time. Only in the case of 'blocker'
transformers - those that are applied on a group-basis - would there
be any sort of meaningful measure; and in most cases you can get this
information from the log file anyway.
Of course I'm not a developer so what do I know? You might find it
turns in up tomorrow's build! But I do think it'd be a tougher
proposition.
Mark
--- In [email protected],
"Jason Birch" <[EMAIL PROTECTED]> wrote:
>
> Mark,
>
> I agree, the feature count is very cool. If we could get the
timings
> for each link too, that would be awesome. It would really help
in
> understanding how different work flows could be optimised.
>
Get the maximum benefit from your FME, FME Objects, or SpatialDirect via our Professional Services team. Visit www.safe.com/services for details.
YAHOO! GROUPS LINKS
|
