anyaevostinar / symbulationemp Goto Github PK

Add a test that figures out where the symbiont/phage offspring will go based on random number seed and checks that they actually go to that location

Delay in loading assets causes weird problems sometimes

When phylogeny tracking is on, Symbulation crashes with a seg fault during the final clean-up. This is due to some messy stuff in the sequencing of the destructors. Basically, what happens is the following:

1. The SymWorld destructor is called.
2. In the SymWorld destructor, we delete the symbiont systematics manager.
3. Once the SymWorld destructor is done, we automatically go into the destructor for its base class (emp::world).
4. The base class destructor goes through the population and deletes each host.
5. The host destructor goes through the host's symbionts and deletes each of them.
6. The symbiont destructor tries to grab the symbiont systematics manager and notify it that this symbiont should be removed.
7. The systematics manager has already been deleted, so this produces a seg fault.

I'm not sure what the best way to fix this is. I came up with some possible solutions, but they're all kind of unpleasant:

• The symbiont systematics manager could get moved to some sort of scope that lives outside the SymWorld
• We could add an extra check in the symbiont destructor that confirms that the systematics manager has not been deleted yet (it's annoying to have to do that check every time a symbiont is deleted just to avoid this one problem at the very end of the run, though)
• We could move the code that notifies the systematics manager that the symbiont has died to SetDead. Is that reliably called every time a Symbiont dies? If so, this is probably the best option.

Catch supports tags as a second argument to TEST_CASE which then allows running just the tests with that tag. These need to be added to all existing tests.

Bad practice, needs to be fixed.

When running symbulation-web.cc, SymSetting.cfg cannot be found. Fixing the file path did not solve the issue.

Add a test that makes sure that when the number of symbionts is at the sym limit, the symbionts in the host vector do not change (not just the number but the actual objects, since there is no ousting allowed)

Several methods in the various organism classes need to be overwritten by subclasses so that the correct type of organism is created, leading to a lot of repeated code that will be difficult to maintain. A function "MakeNew" that does the instantiation of the class will resolve that issue and allow for more inheritance.

The docs:

• don't mention the cookiecutter Github things
• don't mention R studio - add some guidance on installing R
• symbulation vs symbulation-default in simple repeat
• version of emdsk is wrong -- should be 1.38.48
• don't need to cd into web, can use "make serve"
• collecting data has some issues
• creating a new mode should be flushed out
• library API index is malformed
• Add guidance on how the plot generated by ggplot can be interpreted and the big takeaways from the papers so far
• Guide on the settings/things to try next/how to make more changes to them
• What the future/next steps are for code development, what we are hoping people might try adding
• pip3 instead of pip depending on setup

There are raw pointers used throughout that should be replaced with emp::Ptr for safety.

When resetting the petri dish in the web interface, organisms disappear at a different order. Resetting the random seed didn't solve the problem.

Identical if statements for MoveFreeSym in both Symbiont and Phage process. There should be a method created for this so the method can be called instead of repeating code. Existing test for Symbiont process will need to be changed with a new test added for the new method. @K-Johnson-Horrigan can you do this?

There is a great tool for automatically visualizing the layout of git repositories at https://github.com/githubocto/repo-visualizer-demo that we should include in the readme and docs to give a nice overview of the code.

Should change various symbiont containers from std::vector to emp::vector to use the debugging capabilities

The old end_to_end tests are difficult to maintain, need to figure out a better way of testing the main setup code

Every organism should have something useful that they can print out about themselves

PGGHost is lacking in documentation. Each function should be given an input/output/purpose and the instance variables should be given a purpose.

Mutation Rate - slider goes from 0-10 when it should go from 0-1.

Horiz Trans - Docs mention 0 and 1, but the GUI is a switch.

On Windows Chrome, the game mode pops ups show at the bottom of the screen instead of pop up form.

PGGSym is lacking in documentation. Each function should be given an input/output/purpose and the instance variables should be given a purpose.

Change the file name creation in symbulation.cc to automatically include the seed, rather than relying on the user to add the seed to the filename on the command line or in SymSettings

There are several data nodes in SymWorld.h that are lacking documentation. Each function should be given an input/output/purpose.

Organism.h should have documentation. For each function, there should be some documentation that points out the class(es) that the function should be called for.

DistribResources tests currently weird for Host interaction value > 0 and symbiont value < 0

Specifically, making sure symbulation.cc is covered

Because... obviously

Horizontal transmission and vertical transmission mutation rate and size configuration settings currently aren't being tested

Also would be good to move them to their own file to keep things organized