[Nix-dev] Performance impact of Nix

[Marc Weber]

Excerpts from ludo's message of Wed Jul 29 22:24:27 +0200 2009:
> In a recent discussion, Ralf Wildenhues [0] rightfully wondered about
> the performance impact of Nix' file system layout and/or that of the
> symlink indirection in profiles.
No, but I wondered as well. I don't think you can neglect it.
> Did anybody try to measure this?  Thoughts?

Which is the best way to measure it?

Basically you only add one lookup operation so it shouldn't be too much.
One way to measure it is rewriting the env builder to use hard links
instead of symlinks?

How much time does it take to compare paths
/nix/store/32d5rw61pr6ffl0mnbrfs2ih5xjk4nls-vim_configurable-7.2/bin/vim
  instead of
/bin/vim
?

I mean all /nix/store/* paths have at least additional 43 chars in
length and additional 3 directories, don't they ?

So would nix be faster if we used /store only ?

Again: How to measure this ? Reboot a system once?
Compile Open office ?

I'll be on vacation for one week. I can test this when I'm back.

cu
Marc Weber