This reverts commit e8e8164f348a0e8655e1d50a7a404bdc62055f4e. I
misread the original commit as adding the "which" package, but it only
adds it to base.nix. So then the original motivation (making it work
in subshells) doesn't hold. Note that we already have some convenience
aliases that don't work in subshells either (such as "ll").
The openimageio dependency doesn't build on i686. But probably nobody
cares about running Blender on 32-bit anymore.
http://hydra.nixos.org/build/33602734
Note: the 1.2.9 GitHub tarball's sha256 has changed.
Changes:
- Fixed an issue introduced by the fix for 1196 that had the "Upload
to SD" button stop working correctly.
- Fixed an issue causing an error on disconnect after or cancelling of
an SD print, caused by the unsuccessful attempt to record print
recovery data for the file on the printer's SD card.
- Only add bed temperature line to temperature management specific
start gcode in CuraEngine invocation if a bed temperature is actually
set in the slicing profile.
- If a communication timeout occurs during an active resend request,
OctoPrint will now not send an M105 with an increased line number
anymore but repeat the last resent command instead.
- Don't add an extra ok for M28 response.
- Add an extra ok for M29 response, but only if configured such in
"Settings" > "Serial" > "Advanced options" > "Generate additional ok
for M29"
- Trigger M20 only once after finishing uploading to SD
- Prevent M105 "cascade" due to communication timeouts
- Fixed wrong tracking of extruder heating up for M109 Tn commands in
multi-extruder setups.
- Fixed start of SD file uploads not sending an M110.
- Fixed job data not being reset when disconnecting while printing.
Regarding homepage update, changelog says "Moved to pimutils
organization on GitHub. Old links should redirect, but be aware of
client software that doesn't properly handle redirects."
The darwin build has been broken for a long time and I'm unable to
properly debug the issue.
What appears to be happening is that the symbol `HAVE_SANDBOX_INIT` ends
up being defined as 1 while `HAVE_SANDBOX_H` ends up being 0, resulting in
undefined reference errors when `sandbox_init()` is called (because
`<sandbox.h>` is not included first).
This is a regression from dnscrypt-proxy 1.6.0 to 1.6.1.
For context, sandbox.h is a deprecated OSX mechanism for sandboxing.
The build failure is at
https://hydra.nixos.org/build/32705849/nixlog/1/raw
This patch closesNixOS/nixpkgs#14064
Previously, the cisco resolver was used on the theory that it would
provide the best user experience regardless of location. The downsides
of cisco are 1) logging; 2) missing supoprt for DNS security extensions.
The new upstream resolver is located in Holland, supports DNS security,
and *claims* to not log activity. For users outside of Europe, this will
mean reduced performance, but I believe it's a worthy tradeoff.
