testing issue referencing #336
Other test text
LaTeX Test :)
But now I've replied to it and that isn't mentioned here
I don't know whats wrong.
I just invited everyone from the contributers list: https://github.com/CometVisu/CometVisu/graphs/contributors to join this room
but I'm still a bit lost there...
The page is already here (http://cometvisu.github.io/CometVisu/ - domain stuff can be easily fixed later, let it get running first...)
but the point is to get a page that it beautiful but still simple to maintain
so I'm looking at getting Jekyll to run as I hope that it helps
btw, channel #cometvisu on freenode is also extremely quiet...
but I keep it open as "kein Brot frisst" ;)
hey @peuter , quick (we'll see ;-)) question: I'm using mysql to persist my items in openHAB. Strategy is set to onChange, so unlike to rrd i only have records if the item really changed. When showing one of the rather stable items (not changing frequently) this looks a bit odd in the diagram, since often the beginning and end of the curves start/end somewhere in the middle of the diagram. This is because the is no matching event.
I'm wondering how to fix / workaround this issue. Ideally somewhere the whole tech stack could figure out if there is an event preceding the start point of the curve and add a fake event right on the start timestamp. If there is no event past the end date in the persistence storage, the last known value is still the most recent one and a fake entry with that value should be added and the endTimestamp. I hope you get what i have in mind.
The red line cuts out in the middle, reason is that the last value change was at 19:20
Since the strategy is onChance, we know that the value from 19:20 must be still valid for 21:10 (the end of the diagram), otherwise there would be a more recent value
Ideally the system would add a fake event in the result for 21:10 containing the last known value from 19:20
OK now I think I got it. From my point of view, this problem should be solved in the chart library, e.g. draw a flat line to the right end of the chart if there is no data. The data itself should not be manipulated. I am not really familiar with the charting engine used in the CometVisu (I think its flot?), maybe there is some kind of config option which does that.
there is now way to guess the preceding value, which could be anything, drawing a flat line from the first known value to the left is certainly wrong
Adding a fake point somewhere at the beginning is wrong to. I don't see a feasible solution to fill the left side of the chart (and IMHO I don't really see a problem here, ok its not nice but better that faking something which might lead to a chart that does not show the reality).
this should give the the last known value right before the startTime
This probably needs to be handled in the backend on the oh side of things
Yes but then you have at least one dataseries with a value before the requested startTime, wouldn't that lead the chart to augment to the right side and all other series would not not start at the first timestamp? The other problem is that you cannot query openHABs persistance service like an SQL Database it uses an own abstraction layer for the query. I don't know if a query like your suggestion is possible there, might be but not for sure.
not sure if that still works with the svg versions
Where did you add the svgmin task? I can't see a commit on the branch