Re: EventViewer Rewrite

[Phil Muldoon <pmuldoon at redhat dot com>]

Sami Wagiaalla wrote:
> > I'd have to second Phil's questions though - I think we need some
> > semblance of time, or relative time between events on the X axis.
> >   
> The model I am basing this on and the one we already know works is that
> of printf debugging, were the user doesn't really care about time.

Give that particular model and ...
> The design of the new EventViewer makes it really easy to add a time tick
> (just another event). A reasonable suggestion made by Cagney when we
> had this dicussion a while back was to add a time tick every time the 
> event
> loop loops.

The above, there might be a case (at some point) to make the behavior of 
the widget scriptable. (ie time tick or not, just count, etc). 
Personally I can see the printf model working with the tag-sets (tags 
tied to lines of code in a program) but wonder how useful it would to 
know that my program forked in the event-viewer, but not what time it 
forked. The immediate solution to that problem would have a custom 
observer create a log-event and plot to event-viewer. This bring us 
nicely to log and event-viewer integration (ie click a log entry in the 
log view, have it highlighted in the eventviewer). Certainly food for 
thought, and something to think on, down the line.

Regards

Phil