Jacek Sieka

> the scope of the changes really, anything that "already works" but provides infrastructure for multi-platform support - for example the flag changes to `compile`, llvm version handling, etc -...

Also, oddly, the test suite / github actions did not run for this PR - wonder what's up with that..

ok, looks like 523e1736c09a74e0c03d7b63ad5607547cb88232 wasn't enough to make CI run - maybe a rebase would help?

actually, it ran after all, and passes - nice - we'll probably need to differentiate skipped-tests between platforms

the (easy) way to run tests in parallel is to run them as separate processes - it doesn't require anything of the testing framework and makes sure things are nicely...

> the headline feature is parallelism it's not, any more - it has to be enabled, and is not working reliably - there are other features in there however that...

> Have you guys considered spending time improving the asynchttpserver in the stdlib to make it production-ready? It likely won’t involve much effort and would help out the Nim community...

* multiple libraries / applications per repo (monorepo support so that libraries can be published individually) * build c-and-other stuff - `{.compile.}` inside nim doesn't work beyond the most simple...

> but I think we can agree that the presence of a global cache does not prevent us from using Nimble. the cache, as it's implemented today, can't be used...

and for `p2pd` - and ideally, we'd do it for rocksdb as well (rPI instructions include the necessary magic, but we're still vulnerable to distros compiling rocks with different options,...