ste
Forum Replies Created
-
Posts
-
> in the situation where the last turnpoint a pilot take is the
> SS turnpoint, the first crossing of that turnpoint is registered.In any case for the last turnpoint taken we use the first crossing (that gives a valid trail back).
We then look at the remaining tracklog from this crossing to find the point on the log which gives the shortest distance to complete the rest of the task.
If we used the remaining tracklog from the last crossing of the last turnpoint taken we would in some cases not get the shortest distance to complete the rest of the task.
Stein-Tore
From the above follows that,
in the situation where the last turnpoint a pilot take is the SS turnpoint, the first crossing of that turnpoint is registered.
This seems to be the case in one of the samples you sent me where you have 104km task with SS tp having 95km radius.
All pilots take SS tp but not the next one.
What time is registered for the SS tp is not relevant as long as the pilot never take the ES tp.
If you try to widen the radius (to 15km) of the ES tp so that Feri reach it, you should see a different time for the SS tp.
Stein-Tore
Here is an attempt to explain the “logic” of finding a valid trail:
First all crossings of turnpoint cylinders are found which are within open/close periode of each turnpoint. A crossing is the point on the line between two tracklog points where it cross the cylinder wall, does not matter what direction.
NOTE: With big tp radius and error margin, there will be several “crossings” as the cylinder wall may be several 100 meters thick (1000m in the case of tp with 100km radius and 0,5% errormargin) even if the pilot flies straight.
Then we look at the last turnpoint where the pilot had any crossings and where all previous turnpoints also had at least one crossing.
We start with the first crossing of this turnpoint and try to find the crossings of the all previous turnpoints which gives us the fastest valid trail back to the first tp.
If we do not find a valid trail we repeate the process with the next crossing for this turnpoint …If the turnpoint does not have any crossings which will give a valid trail back to first tp we repeate the process with the previous tp …
This means that for the set of turnpoints in the valid trail the first possible crossing for the last turnpoint is used and latest possible crossings is used for all the previous turnpoints.
If the end of speedsection (ES) turnpoint is not the last turnpoint in the valid trail we find the first possible crossing for this turnpoint.
This should give the fastest valid trail for the pilot.
The general idéa is to find the last possible crossing of the SS turnpoint which leads to the first possible crossing of the ES turnpoint, hence giving the fastest valid trail between SS tp and ES tp.
If the ES turnpoint has not been reached the actual crossing times are fairly irrellevant for scoring.
Stein-Tore
This is just a new way of limiting an elapsed time task. Normal way is to set close for task (or tps) at a given time of the day.
You need to set it relative to the pilot’s start-time.
Would it be ok to use the last time a pilot leave the (first) turnpoint as start-time?
The task definition would then be either “Elapsed time” or “Open distance”.
The addition will simply be a field for spesifying max elapsed task time.
The difference from what you’re asking is that we do not attempt to find out when the pilot leaves the ground. So the pilot is free to start the task at any time once of the ground (just like any ordinary elapsed time task).
Stein-Tore
Looking into it, seems the specifications is not fixed yet, only a draft in the 2009 rules and this is not very well described either.
If anyone can clarify it would be good.
Stein-Tore
Stein-Tore
An updated version of FS 1.2.12 is at http://fs.fai.org/wiki/CurrentVersion now which fix some special cases regarding tasks and wpt coordinates.
FS 1.2.13 is just around the corner and should handle any sort of weird tasks and will also have some improvments in displaying tasks & tracklogs and airspaces (hopefully).
Stein-Tore
See GAP2007A at http://fs.fai.org/wiki/ScoringFormulas …
Stein-Tore
I agree.
It is actually not related to “1000 pts …” option.
The time points punishment happends regardless and it does look a bit strange having the whole field getting sign. less points when the winning pilot just happends to lands btw ES tp and goal tp as opposed to when the winning pilot lands just before ES tp (or makes goal).
Will make a revision of GAP2007 named GAP2007A where the time points punishment will only be used in cases where at least one pilot reach goal tp.
Stein-Tore
A question regarding this:
In FS any scoring formula basically works with the information present in the FsTaskDefinition element and FsFlightData elements.
So in this case I’ve added a new optional attribute, “start_gate”, to the FsFlightData element.
Then it is upto the scoring formula to use this or not.
As it is now, this is a feature of a specific scoring formula (OZGAP2005V1).
Should this sort of information be regarded as a feature of the scoring formula (as I’ve made it now)?
Or should any scoring formula use this information (“start_gate” attribute of FsFlightData element) if it is present?
Stein-Tore
See http://fs.fai.org/ticket/146 …
Stein-Tore
With current version of FS:
Since it is only 20 pilots I suppose it is possible to check these manually.
I suppose most of them might get the correct mandatory start gate by FS.
Then for those not getting the mandatory start gate you need to take some action.
Those starting too early should either get 0 distance or some penalty (jump the gun…).
For those taking a later start gate I suppose you can just change the start time manually to the mandatory start gate time.
Stein-Tore
Hi Tom,
could you zip up the files and add it to ticket 119.
Version 1.2.8 will be out shortly with a fix for 119 but I would like to test it with your data before I release it.
Stein-Tore
