catchouli / coment

A C++11 header-only ECS (entity-component-system) library (no longer updated but feel free to use or extend it)

Hi there 👋

My name is Caitlin and I'm a senior engineer from the UK.

A few of my specialities are:

• Rust, C++, C#, Haskell (but I'm known for being able to use just about any language :P)
• High performance applications such as 3d graphics, GIS and data processing
• Docker and microservices architectures

I'm also a game developer in my spare time, (and anything else I think of! check out my repositories page 😉), and love figuring out how to build new things.

I have a github organization I use to (somewhat unsuccessfully) stop my repositories list from getting out of control: https://github.com/catchouli-old

See https://cat.bio/ for my full profile 👍

coment's Issues

Shared library compile option

It would be nice to be able to compile comet as a shared library rather than just a static library.

sudo make install include files

Please mistake me if I'm wrong, but it doesn't look like 'sudo make install' is copying the header files to /usr/include[/coment]. This should be a simple addition of something like 'cp -r $(INCDIR) $(INSTALL_DIR_INC)'

Arch Linux PKGBUILD file

Not sure where general discussion for this project should be, so I've just created this issue to notify anyone that I have made a PKGBUILD for those that wish to easily install this library under arch linux. If there is a better place to discuss this (e.g. a forum somewhere) please let me know for future anouncements.
Anyway, it is hosted here: https://aur.archlinux.org/packages/coment-git/

Utilise $(DESTDIR) and $(PREFIX) for better cross distro support.

To allow support from more distros, it may be worth reconstructing the lines:
INSTALL_DIR_LIB = /usr/lib
INSTALL_DIR_INC = /usr/include
to instead read:
INSTALL_DIR_LIB = $(DESTDIR)$(PREFIX)/lib
INSTALL_DIR_INC = $(DESTDIR)$(PREFIX)/include
This would allow for much better support for distros that may wish to eventually package your library (I was gonna make a PKGBUILD for arch linux, but it's not possible without this change first)

It might be time for some branches

I think it could be time to start organising this project into multiple branches rather than cramming everything into master. Comet is looking more like a successful OSS project each day, and so I believe we should create a good branch model to help promote development. Something like this would be good: http://nxvl.blogspot.com.au/2012/07/a-continous-delivery-git-branching-model.html
Already we are seeing awesome features being developed that aren't being accepted upstream. At least if we use something along the lines of whats been referenced above, those features can sit in the upstream repo to be merged into master at a later date.
I might be completely out of my league here, please let me know what your opinions are.

Bitset class

Although efficient, the bitmasks used for systems, components, groups and tags are limited to a compile time length of 32. This means that to have more than 32 of any of these things, recompiling the library would be necessary. This has long made me worry about the scalability of the library.

It wouldn't be too difficult to make bitsets which automatically grow, given that the size is already specified at compile time. I think the reason I didn't originally do this was for efficiency concerns, but I believe that this is premature optimisation.

Different types of systems

Currently the only default type of user system is the EntityProcessingSystem, which processes one entity at a time based on a set of included (and excluded, once I've gotten around to merging jpschorr's pull request.) components. Artemis also has another set of components, of which any will be matched. I'm intending to implement this at some point unless somebody else beats me to it.

What other types of systems would be useful to include by default? For example:

• A system which processes entities at a specified interval of time. I'm thinking that you'd still call update on it every frame, but once that specified interval of time has passed it'll process.

I've also had requests for systems that process one set of entities against another, and entity collections that stay up to date like a system but without any attached logic.

Does anybody else have any thoughts on this subject?

Fast balls in Sample

It seems that when the user pauses then unpauses the sample (using 'm'), there is a dratic change in the velocity. This isn't a particularly urgent issue as it only affects the sample, but it would be more aesthetically pleasing to be able to demonstrate the framework using a sample that functions as intended.