previously, stderr was garbled, which made it really annoying to deal
with build errors.
it turns out that the library wrapper around calling a process is not
just counterproductive, but also completely unnecessary.
the build now takes longer because it's not incremental any more.
but the problem with `sphinx-build` in a shell environment is that it
doesn't keep track of an entire class of files, that would have to be
considered manually. this is too much manual overhead that doesn't scale
at all. with the new setup the live preview will do exactly what the
final deployment does.
[1]: https://github.com/executablebooks/sphinx-autobuild#relevant-sphinx-bugs
Co-authored-by: Alex Groleau source@proof.construction
* mv question -f faqs -t nix-recipes
* add nix-recipes to page index and fix hierarchy
* change question to statement
* rephrase a few sentences and add links
* move the question to troubleshooting
* fix broken link
---------
Co-authored-by: Valentin Gagarin <valentin.gagarin@tweag.io>
* shorten the sharing dependencies article to a guide
the contents do not really warrant a full-blown tutorial
* use an example with an explicit build-time dependency
* fix typo
* link `inputsFrom` to Nixpkgs manual
The current code has a number of issues.
1. realising the code results in:
``
markers=$(geocode 'New York') size=640x640 scale=2 zoom=7 ...
^-------------------^ SC2046 (warning): Quote this to prevent
word splitting.
``
2.
``
$(geocode ${lib.escapeShellArg marker.location})"
``
Assumes that geocode has been added to the PATH so replace with
``
${config.scripts.geocode}/bin/geocode
``
as per previous examples.
3.
``
in "markers=${ lib.concatStringsSep "\\|" attributes}";
in builtins.map paramForMarker config.map.markers;
``
The ordering of this actually creates multiple ``[markers =]``
parameters and does not use the concatenation string to join the
attributes together (in effect there is only 1 attribute per call) to
map, and 1 new ``marker`` created.
Signed-off-by: Brian McGillion <bmg.avoin@gmail.com>
* tutorials: fix passing paramaters to geocode
When running the package as-is the following error occurs
``geocode: line 7: $1: unbound``
This is caused because ``scripts.geocode`` does not pass the parameters
down to the wrapped script ``geocode``.
Signed-off-by: Brian McGillion <bmg.avoin@gmail.com>
* Draft module system introduction tutorial
* add intro/conclusion, rework prose, follow styleguide, clean diffs
* Review pass
* Apply suggestions from code review
* fix whitespace
* add some more motivation the each section
* make scripts downloadable
* address review comments
* make script actually work
...hopefully. can't test it without Google API key
* add file watching
yes, this looks scary, and yes, it works.
* update diff and wording
* more notes on potential pitfalls
* be explicit which `map` we mean
* split nullable from default values
* also wrap the geocode script
* work through the tutorial to the end
* add tutorial overview
* `lib` is always passed
* add separate section for `evalModules` and fix link
* make option strucutre more self-explanatory
* explain command line invocations
* add note on incomplete reference documentation
* add more highlight to the `config` distinction
* fix parameter passing to the `./map` script
* fix typo
* fix wording
* link to summer of nix
* add missing word
* link to Google Maps API docs
* more explicit requirement
* use correct module system terminology
* Update source/tutorials/module-system/module-system.md
* Apply suggestions from code review
* whitespace
* module-system.md: replace comments with captions
* add missing lang for code-block
* Update module system title
* change most headers to be about module features (#797)
* change most headers to be about module features
Some headers could not be made about module features, and that's a
strong signal that those sections should be removed.
* Apply suggestions from code review
* module-system.md: Fix header casing
Co-authored-by: Alexander Groleau <source@proof.construction>
Co-authored-by: asymmetric <lorenzo@mailbox.org>
Co-authored-by: Valentin Gagarin <valentin.gagarin@tweag.io>
* add SOURCE_DATE_EPOCH to Makefile to fix copyright year
* use date to get unix time in place of git log
* Update Makefile
pass in --date='Jan 1' to date as suggested
Co-authored-by: Silvan Mosberger <github@infinisil.com>
---------
Co-authored-by: Silvan Mosberger <github@infinisil.com>
recent additions to the Nix manual clarify the distinction between
search path and lookup path, and also document the resolution algorithm.
lookup paths are now a distinct Nix language construct with its own
reference documentation.
* Remove specific software name from header
There's enough detail about it below, and the specific tech isn't the point of this section anyway.
Co-authored-by: Valentin Gagarin <valentin.gagarin@tweag.io>
* rewrite guide to pinning dependencies with niv
* use the specific branch directly
Co-authored-by: Silvan Mosberger <github@infinisil.com>
* use the shell's niv explicitly
* Apply suggestions from code review
---------
Co-authored-by: Silvan Mosberger <github@infinisil.com>
this removes a difference to how Nix currently fetches flake references.
using a constant is preferrable because renaming the source won't force
downstream rebuilds of store objects cached from when flakes were used.
upstream fix: https://github.com/nmattia/niv/pull/379
use niv for dependency management.
this follows our own tutorials on writing packages and development
shells, and we generally find it is easier to read and work with.
Co-authored-by: Alejandro Sanchez Medina <alejandrosanchzmedina@gmail.com>