| Age | Commit message (Collapse) | Author |
|---|
|
Add response body to network errors
|
|
obsidiansystems/update-chunksize-to-suggested-value
Update chunkSize to the suggested value
|
|
This was a suggested course of action in a review in one of our earlier
commits, https://github.com/NixOS/nix/pull/3801#discussion_r457557079
|
|
|
|
Constant space `addToStoreFromDump` and deduplicate code
|
|
README: improve development docs
|
|
Correct bug, thoroughly document addToStoreSlow
|
|
github.com:obsidiansystems/nix into from-dump-stream
|
|
|
|
Co-authored-by: Eelco Dolstra <edolstra@gmail.com>
|
|
Due to https://github.com/NixOS/nix/issues/3841 we don't know how print
different messages for different verbosity levels.
|
|
|
|
Optimize `addToStoreSlow` and remove `TeeParseSink`
|
|
Co-authored-by: Eelco Dolstra <edolstra@gmail.com>
|
|
|
|
Add newline to profile sourcing line #3393
|
|
|
|
fix make's impurity on /bin/sh
|
|
nix edit: call restoreSignals() before `execvp`-ing the $EDITOR
|
|
This is important when using tooling like BEAR to generate
compilation database since the used glibc version needs to match
for LD_PRELOAD to work. It might be also beneficial when building
on systems other than NixOS with nix develop since /bin/sh might
be not bash (which is what all nix devs use for testing).
This fix is not perfect because Makefile.config.in itself is
also build with make but strictly better than the status quo.
|
|
Currently resizing of the terminal doesn't play nicely with
nix edit when using kakoune as the editor, as it relies on the
SIGWINCH signal which is trapped by nix. How this is not a problem
with e.g. vim is beyond me.
Virtually all other exec* calls are following a call to
restoreSignals(). This commit adds this behavior to nix edit
as well.
|
|
Add back flake-compat shell.nix
|
|
This was removed in the merge commit
adf2fbbdc2c94644b0d1023d844c7dc0e485a20f. I think this was a mistake
that occurred when resolving a conflict.
|
|
|
|
|
|
|
|
This shows all changes between generations of a profile. E.g.
$ nix profile diff-closures --profile /nix/var/nix/profiles/system
Generation 654 -> 655:
nix: 2.4pre20200617_5d69bbf → 2.4pre20200701_6ff9aa8, +42.2 KiB
Generation 655 -> 656:
blender-bin: 2.83.0 → 2.83.1, -294.2 KiB
Generation 656 -> 657:
curl: 7.68.0 → 7.70.0, +19.1 KiB
firmware-linux-nonfree: 2020-01-22 → 2020-05-19, +30827.7 KiB
ibus: -21.8 KiB
initrd-linux: 5.4.46 → 5.4.49
...
|
|
We use this to simplify `LocalStore::addToStoreFromDump`.
Also, hope I fixed build error with old clang (used in Darwin CI).
|
|
|
|
|
|
|
|
|
|
|
|
This reverts commit 592851fb67cd15807109d6f65fb81f6af89af966. We don't
need this extra feature anymore
|
|
I got it to just become `LocalStore::addToStoreFromDump`, cleanly taking
a store and then doing nothing too fancy with it.
`LocalStore::addToStore(...Path...)` is now just a simple wrapper with a
bare-bones sinkToSource of the right dump command.
|
|
|
|
This reverts commit cff2157185912025c24a1b9dc99056161634176c.
|
|
from-dump-stream
|
|
This was completely broken since d8972317fc4314864619cadd5620ae780da657a3.
|
|
|
|
For instance, 'nix why-depends --use-derivation nixpkgs#hello
nixpkgs#glibc' shows why hello's .drv depends on glibc's .drv.
|
|
|
|
|
|
That is, the commands 'nix path-info nixpkgs#hello' and 'nix path-info
/nix/store/00ls0qi49qkqpqblmvz5s1ajl3gc63lr-hello-2.10.drv' now do the
same thing (i.e. build the derivation and operate on the output store
path, rather than the .drv path).
|
|
|
|
This command makes it easier to see what changed between two closures,
i.e. what packages/versions got added or removed, and whether there
were any notable changes in path size.
For example:
$ nix diff-closures /nix/var/nix/profiles/system-655-link /nix/var/nix/profiles/system-658-link
blender-bin: 2.83.0 → 2.83.2, -294.2 KiB
curl: 7.68.0 → 7.70.0, +19.1 KiB
firmware-linux-nonfree: 2020-01-22 → 2020-05-19, +30827.7 KiB
ibus: -21.8 KiB
initrd-linux: 5.4.46 → 5.4.51, +16.9 KiB
libexif: 0.6.21 → 0.6.22, +497.6 KiB
linux: 5.4.46 → 5.4.51, +13.2 KiB
mesa: 19.3.3 → 19.3.5, -183.9 KiB
nix: 2.4pre20200701_6ff9aa8 → 2.4pre20200708_9223603, +9.7 KiB
nix-bash-completions: 0.6.8 → ∅, -57.6 KiB
nixos-system-hagbard: 20.03.20200615.a84b797 → 20.03.20200713.add5529
nvidia-persistenced: 440.82 → 440.100
nvidia-settings: 440.82 → 440.100
nvidia-x11: 440.82-5.4.46 → 440.100-5.4.51, +664.7 KiB
pcre: 8.43 → 8.44
php: 7.3.16 → 7.3.20, -26.2 KiB
python3.7-youtube-dl: 2020.06.06 → 2020.06.16.1, +8.4 KiB
samba: 4.11.5 → 4.11.9, +30.1 KiB
sane-backends: 1.0.28 → 1.0.30, +680.5 KiB
source: -182.0 KiB
zfs-kernel: 0.8.3-5.4.46 → 0.8.4-5.4.51, +9.9 KiB
zfs-user: 0.8.3 → 0.8.4, +20.1 KiB
|
|
This reverts commit a2c27022e9afc394e08d34d349587c8903fc1a97. See
addToStoreSlow(), we don't need to handle this case efficiently
anymore. In fact, we can almost remove the method/hashAlgo arguments
since the non-recursive and/or non-SHA256 are almost not used anymore.
|
|
|
|
We were calculating the nar hash wrong when the file ingestion method
was flat. I don't think there's anything we can do in that case but dump
the file again, so that's what I do.
As an optomization, we again could reuse the original dump for just the
recursive and non-sha256 case, but I rather do that after this fix, and
after my other PRs which deduplicate this code.
|
|
Rather than copying byte-by-byte, we let the coroutine know how much
data we would like it to send back to us.
|
