[Nix-dev] Re: [Nix-commits] SVN commit: nix - r25375 - in

[Eelco Dolstra]

Hi,

On 01/05/2011 11:16 AM, 7c6f434c at mail.ru wrote:

>> I've disabled this in all-packages.nix because the fetchgitrevision call breaks
>> "nix-env -q".  This is because "nix-env -q" won't perform derivations during
>> evaluation.
> 
> Is it because name attribute cannot be evaluated, or would it fail the same way
> for anything that affects the output path?

For "nix-env -qa", it's because the name or system attribute cannot be
evaluated.  For "nix-env -qa --drv-path" (or similar), anything that affects the
drvPath would cause it to fail.

> Should there be a switch in Nix
> to break such tricks even for non-readonly evaluation?

Maybe.  For Hydra, it would also be useful to disable builtins.currentSystem and
builtins.getEnv.

>> Also, perhaps these packages should be kept in a separate tree of "bleeding
>> edge" developer/maintainer-targeted Nix expressions, since Nixpkgs should have
>> at least some minimum level of stability :-)
> 
> Well, stability is not the right word here, I am afraid.. "nixos-rebuild switch" failing
> after a successful "nixos-rebuild build" because of a package would destroy our main
> selling point

Yup, this is why "builtins.currentTime" is an undocumented "feature" which
really should not be used in Nixpkgs :-)

-- 
Eelco Dolstra | http://www.st.ewi.tudelft.nl/~dolstra/