Due to improper sandboxing, I didn't notice the failed tests while
upgrading jupytext in #121648. The errors come from jupyter_core so
the same fix is used here as there: set $HOME to a writable directory.
Guile's version is part of the path and the installed files won't be
found if we don't include it. Also, we can rely on the paths of build
inputs being added to $GUILE_LOAD_PATH and $GUILE_LOAD_COMPILED_PATH
by Guile's setup hook.
pytest-sanic is incompatible with the current version of Sanic, see
sanic-org/sanic#2095 and yunstanford/pytest-sanic#50. While it is
broken, we also need it to run Sanic's tests (for which case it works
just fine).
Previously we used nodePackages only to fetch Iosevka's build
dependencies and then fetched the source code ourselves. Updates
involved changing the version and hashes in the `iosevka` derivation and
then running node-packages/generate.sh to update the build dependencies,
which in turns updates *all* of node-packages.nix.
A new proposed policy for handling node-packages.nix updates involves
batching those updates. Previously, that would mean `iosevka` and its
build dependencies could end up out of sync until the batched update was
run.
To work with the new policy, we now fetch Iosevka's source code (and
not only its dependencies) through nodePackages. Updates are done by
changing the source URL in node-packages.json, which eventually becomes
part of an update of node-packages.nix, which is then propagated to
`iosevka` itself.
One con of this strategy is that errors can not be caught directly
after the update, but only after node-packages.nix is regenerated.
Builds currently fail with `ar` trying to operate on what are clearly
two paths concatenated together. It stems from a backward-incompatible
change in Make:
> Previously appending using '+=' to an empty variable would result in
> a value starting with a space. Now the initial space is only added
> if the variable already contains some value. Similarly, appending an
> empty string does not add a trailing space.
This issue was first reported on the MAME repository proper
(https://github.com/mamedev/mame/issues/6248), and affects libretro's
2016 snapshot as well. A fix that is reported to work with previous
versions of Make was upstreamed to:
- GENie, the build system: https://github.com/bkaradzic/GENie/pull/493
- MAME: https://github.com/mamedev/mame/pull/6262
- libretro: https://github.com/libretro/mame2016-libretro/pull/47
The fetched patch comes from the last of these.
The TUI library aerc uses, tcell, comes with a built-in database of
terminal definitions. For terminals not in this database (e.g.
Alacritty), an entry is generated dynamically using infocmp, part of
ncurses' dev output.
