-
Notifications
You must be signed in to change notification settings - Fork 69
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Unexpected .sfs growth with #3200 test build #3203
Comments
1 more month of timewarp, the 140MB has already gone back up to 185MB. Definitely something off. |
I read the first save and put a few traces to see what is happening. There are many histories with ~700k points, which seems like a lot, but still some downsampling takes place. More analysis will be needed.
|
Each of the |
As requested: JOURNAL.zip (~150MB) |
Thanks for the journal, it greatly sped up tracking down the bug. The explanation is that Principia/physics/discrete_trajectory_segment_body.hpp Lines 436 to 439 in df1e692
is an incorrect translation of Principia/physics/discrete_trajectory_body.hpp Lines 682 to 684 in 3473dbc
In the old code, control then falls through to the code that actually does the downsampling. In the new code, if the entire dense segment could be fit by a single polynomial within the tolerances (as opposed to multiple polynomials), that is, if the trajectory is very smooth… we just don’t use that fit, we leave those ten thousand points lying around and move on. |
Fixed by #3204. |
(opening this as requested on discord, mostly as a place to put links to saves)
Seen while running test3200 build from https://discord.com/channels/319857228905447436/480397772248580098/909198803633053706
Unclear if it's related; growth seems much faster than before, but I had also just launched an extra 10 vessels.
0 - 1961.zip
The original 170MB .sfs
wip.zip
The 140MB .sfs after Hack Gravity trick
I later tried deleting the in-orbit vessels altogether; no meaningful difference. Deleting the oldest interplanetary vessel (0155 uranus 2, 1190d old) shaved off 4MB, don't know if that's within expectations. Deleting the newest (0291 pluto 3-2, 81d old) shaved off ~200KB, not obviously wrong either
The text was updated successfully, but these errors were encountered: