Lester Hedges

Hmm, but if you delete the frames from the underlying system object, then it doesn't work as expected again: ```python import sire as sr mols = sr.load_test_files("ala.top", "ala.crd") mols =...

Yeah, there's weirdness going on. For example, if you delete a single frame (the first frame in this case, but it doesn't matter): ```python import sire as sr mols =...

This is what happens when I print out the call to `System::saveFrame`: ``` System::saveFrame System::saveFrame System::saveFrame System::saveFrame System::saveFrame mols currently has 5 frames System::saveFrame System::saveFrame System::saveFrame System::saveFrame System::saveFrame mols currently...

Ah, I see, there's a `SystemFrames` object that stores snapshots from the live trajectory. I think this isn't in sync with the trajectory property in the system somehow. In any...

Yes, I've updated `somd2` and it works perfectly. Can revisit this at a later date if needed.

Great. This is what I'm now doing, i.e. writing the last N frames as a chunk at a checkpoint, then assembling the whole trajectory at the end. I guess I...

The random crash on interpreter exit is caused by the [CacheData::cleanUpOnExit](https://github.com/OpenBioSim/sire/blob/4fcc147db1afa4ff5c8e868484eaa18eda3d4ff2/corelib/src/libs/SireBase/pagecache.cpp#L618-L673) method. If I comment this out, then I see no crashes whatsoever. I'll take a look to see if...

Okay, it's _entirely_ caused by the final bit of the block [here](https://github.com/OpenBioSim/sire/blob/4fcc147db1afa4ff5c8e868484eaa18eda3d4ff2/corelib/src/libs/SireBase/pagecache.cpp#L657-L672) where it waits for the threads to stop, then kills them if they don't. It looks like this...

That's a good idea, I'll do that.

Another update on this. I've checked what's going on in the `SystemFrames` code and even when calling `deleteAllFrames` on the system, the underlying frames object is not cleared. For example,...