Robert Hensing

I was thinking something along the lines of ```nix attrsWith { name = "username"; itemType = submoduleWith { modules = [ ]; }; # or, perhaps later: # itemTypeFunction =...

> Should `` be configured via parameter `name` ? This should be considered in the context of also having e.g. https://github.com/NixOS/nixpkgs/pull/218812/files, which is a laterally related but independent parameter. This...

Could we perhaps merge the changes that are mere refactors first? Unfortunately the first commit already introduces a new feature. Just having a very basic `attrsWith { lazy? }` merged...

Assertions have some conflicting requirements: - For performance, we want attributes, options values and definitions to be as lazy as possible. - For the best relevance of error messages, we...

I suppose you could, but it does duplicate the whole module evaluation. What is the benefit over, say, triggering the checks as part of a particular attribute, such as an...

Finally fixed in https://github.com/NixOS/nix/pull/11009

Also affects nix-darwin - https://github.com/LnL7/nix-darwin/issues/970

Came across this in this new test https://github.com/NixOS/nix/pull/8566/files#diff-79ff5e5f6f39edd4a8a247ceb45dddde66f90d2c93ae89ebf37e16f69d0e8df4R160 Probably that has the same cause, and this may be an easy fix.

> We could also get rid of the ambiguity by dropping the integer datatype. So `4` would be a floating point number exactly equal to the integer 4. This would...

I think this issue can be solved by improving `Value::print`, in a way that always outputs a decimal point when rendering a float. The method is not used in a...