The Future of FS?
Vitaly Umansky writes in the BaseCamp group that follows FS:
Hello everyone. I was recently provided access to FSComp sources
and finally had some time make a deep dive study and implemented a few things
there. As basecamp doesn't appear to dsiplay inline images nicely, and also as
this is fairly long to be a post, I've made kind of mini article with my report,
which is linked to below.
FSComp appears to be used in many competitions and effectively is the gold
standard for now even with all its deficiencies. From reading previous topics in
this basecamp group it's not quite clear what's its future is, i.e. what are the
plans. Are we moving to Naviter software (I've heard, looks like not), or PWC
CompCheck. What's going on with FSComp, is it going to be open sourced or just
abandoned? Is it going to be further developed and supported? What's the
official position on all these issues?
There are quite few things that can be implemented and improved, some of the UI
and computation improvements I've made already (locally yet), such as tolerance
bands visualization, computation and visualization of true WGS84 and FAI sphere
cylinders (currently UTM is used for visualization), some fixes in speed section
length computation and leading points calculation, and quite few more in
UI. There are also few more things I'm still working on. It would be nice to
know what the vision and plans are for future development, so that my time spent
on making improvements would be felt as usefully spent in the end.
Naviter`s SeeYou is great, you could almost say it is ready, but from what I
have talked with Andrej, GAP is the major problem of SeeYou. Every time, they
compare it with FS, there is always some difference. SeeYou is great if we would
like to come up with a new scoring formula, ready to use if we would like to
test it, or use it already for Time Based scoring, everything is there. BUT,
when I mention GAP to Andrej, it is almost a dead end.
So from my point of view, FS, is something we (I) would like to keep alive. Most
(or a lot of) of the cat2 organizers are using it, and it is a great service
from us to the pilots. So if someone is willing to dig into it and improve (if
that is actually possible), then by all means, lets give it a try.
Leading points calculation
Speed section distance is defined Section 7F as "the path of shortest distance
from start of speed section to end of speed section that touches all turnpoint
cylinders". Currently it's computed as length of subpath between SSS and ESS
cylinders of optimal full task route (path between takeoff cylinder and goal
cylinder), which is not correct and differs significantly (see Fig. 6). Correct
SS distance computation is essential for leading points computation and
difference can be way more than 10 pts (on a task with around 150.0 pts for
leading are available).
During leading coefficient computation, first member of time graph is omitted.
This may be due to the fact that ESS distance is computed incorrectly, thus once
pilot crosses SSS, it may be computed that pilot has own already a few
kilometers in SS, and once things get squared, that addition is huge, and makes
ridiculous changes in LC. With proper SS distance computation it's not possible
and first member can be used (and it does make difference, as expected in the
beginning of the task). For some reasons power 0.666 is used instead of 2/3.
This makes small changes (like 0.1 pts for leading points score), but due to
rounding it can easily change resulting score and ranking. There is no need for
The future of FS is being discussed in a basecamp forum among a
few participants some of whom have access to the code. Vitaly is one of the
participants and there are others who have other ideas. Andrej from Naviter is
involved also and so is Joerg Ewald, who currently maintains the code. It's
great to see multiple voices and people willing to make changes to improve FS or