nanofil/survey/default.nix
The nanofil package survey (docs/survey.md). nanofilize is a package adapter: drv built with nanofil as its C compiler, in the Fil-C staging package set (whose libraries, built by filcc, it links with), and nothing else compiling its C: cc.sh hands every C compile to nanofil-cc and refuses what it cannot, so a package builds with nanofil alone or fails. C++ still goes to filcc and is counted (cxx.sh). Its own checkPhase runs where the port's package runs one.
nix build -f nanofil/survey ref.zlib # the filcc build nix build -f nanofil/survey nf.zlib # the same package by nanofil nix eval --json -f nanofil/survey meta.zlib # versions, sources, deps
The build log ends with "nanofil-survey: N objects, M C++, K fallbacks" on success, and on failure also the first nanofil errors that configure, meson or cmake logged.
| repo | default ../.. |
nanofil options for every compile, e.g. "-O" ( | |
| opt | default "" |
Only bin/ and src/, so that the survey's own results and docs do not change the nanofil builds.
A package's own build dependencies' names, to set aside what needs LLVM, Rust or Go.
Packages whose builds measure something other than C compilation, or regularly need more than the survey's 90-minute limit. The default survey sets them aside; campaignctlsubmit bench builds them, one at a time on swa, as a compiler-speed benchmark (docs/campaign-service.md).