A number of recent additions have only been half-implemented and probably shouldn't be sitting around on master, but there they are. The only real benefit to this over feature branches is that I don't have to worry about difficult merges after a refactor that moves a lot of files around. The downsides (beyond the usual mayhem and confusion) are tons of warnings and broken test suites, some of which I've even had to disable just so that they don't stop the other test suites from running.
these include
[x] symmetry analysis in rsp2-structure (which is incomplete)
[x] band unfolding in rsp2-tasks (which is entirely broken and hellishly overcomplicated, and is currently the source of many unused code warnings)
[x] cell reduction in rsp2-structure (which is untested)
[x] hnf searching and/or reduction (there's a fairly optimized 2x2 search in garbage but we want 3x3)
[x] Source/Save/Load in rsp2-tasks (I want to start using these all around the crate)
(Like the garbage directory wasn't bad enough...)
A number of recent additions have only been half-implemented and probably shouldn't be sitting around on master, but there they are. The only real benefit to this over feature branches is that I don't have to worry about difficult merges after a refactor that moves a lot of files around. The downsides (beyond the usual mayhem and confusion) are tons of warnings and broken test suites, some of which I've even had to disable just so that they don't stop the other test suites from running.
these include
rsp2-structure
(which is incomplete)rsp2-tasks
(which is entirely broken and hellishly overcomplicated, and is currently the source of many unused code warnings)rsp2-structure
(which is untested)garbage
but we want 3x3)Source/Save/Load
inrsp2-tasks
(I want to start using these all around the crate)