Pascal Leroy

Vessel destruction is expensive

1

While replaying the journal for #3375, I am noticing that destroying a vessel is rather expensive (in some cases, 20-30 seconds). I have not investigated why, but I suspect that...

Most papers use `x, y, yʹ` but our APIs generally use `y, yʹ, x`. The 1st-order equations have been corrected already.

Investigate the possibility to display equipotential lines

2

What I have in mind is [something like this](https://static.scientificamerican.com/sciam/cache/file/799B2D41-709C-4511-9D60A3E4B946DB55_source.png?w=590&h=800&5081446F-BF42-45E4-89D85528A2C70365). There are many interesting challenges, though, in particular (1) how insanely costly is that going to be? (2) what kind of...

Downsampling is slow

7

There has to be a better way to approximate a trajectory by a low-degree polynomial. The current implementation does *O*(N Ln N) evaluations of the cubic sequentially.

https://github.com/mockingbirdnest/Principia/blob/48f2466671bb206fc0999da38f36c1eb3b6a6f80/ksp_plugin_adapter/ksp_plugin_adapter.cs#L1337-L1339

See the discussion in #2550, it should make it possible to declare the type of the flag and probably its default value.

In Frobenius, only solid-state boosters lose angular momentum. For liquid-fuel engines it's a reasonable phenomenology to assume that the fuel goes through an infinitely thin conduit on the axis of...

Keep track of the vessels when they are managed by KSP

3

This is a follow-up on #2443, the solution of which is a bit of a hack ("zombies"). We could keep the `Vessel` objects at all times and observe what KSP...

From the forum: > Is there anyway to have the RPM (IVA) Navball change in the same way as the game navball? > Ideally I would like to use exclusively...

Uniform use of FORCE_INLINE

2

Apparently, for namespace-level functions, it needs to go on the definition. For static members of classes, it can go on the declaration. To make things worse, C4714 seems to be...