| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
| |
Those tell the browser that it’s going to need them later, even it
hasn’t found them yet (e.g. the fonts can only be found after loading
the CSS).
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
From the release notes of Git version 2.27.0:
* "git pull" issues a warning message until the pull.rebase
configuration variable is explicitly given, which some existing
users may find annoying---those who prefer not to rebase need to
set the variable to false to squelch the warning.
This is exactly the warning which is annoying me all the time now, so in
order to get rid of it, let's explicitly set the default behaviour
(which is doing a recursive merge).
Just to be sure that I really want the default behaviour, I analysed my
shell history for invocations of "git pull" and only around 20% of the
invocations were with --rebase, 14% were with an explicit URL (but no
rebase) and the rest were recursive merges.
Signed-off-by: aszlig <aszlig@nix.build>
|
|
|
|
| |
This seems useful.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Both are the enhanced edition from GOG. Packaging is mostly
straightforward, although I had to use patchelf with a custom patch,
since while --remove-needed removes the DT_NEEDED entries from the ELF
it doesn't however remove the corresponding entries in the
.gnu.version_r section.
The reason why I did this is because we really should not need Expat and
OpenSSL, because they're only used by the XMPP portions of the
statically linked libjingle.
Signed-off-by: aszlig <aszlig@nix.build>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
I'm not familiar with the "easy-dhall-nix" project, but the repository
is imported into Vuizvui via import-from-derivation. While this by
itself is not a big issue (apart from contributing hugely to evaluation
time, we're already at around an hour), the "dhall-nix" derivation is in
turn imported again via importDhall, so whenever something breaks with
dhall-nix, our evaluation will break as well.
Unfortunately, something is broken right now:
building '/nix/store/c363947v9qk287d07qj2kpj60rmzwalj-dhall-nix-1.1.14-x86_64-linux.tar.bz2.drv'...
trying https://github.com/dhall-lang/dhall-haskell/releases/download/1.32.0/dhall-nix-1.1.14-x86_64-linux.tar.bz2
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 648 100 648 0 0 2592 0 --:--:-- --:--:-- --:--:-- 2581
100 2255k 100 2255k 0 0 1639k 0 0:00:01 0:00:01 --:--:-- 6287k
hash mismatch in fixed-output derivation '/nix/store/yhls1ffnvp1nbjsm0xr3l1z6j6x4waqh-dhall-nix-1.1.14-x86_64-linux.tar.bz2':
wanted: sha256:1j32jf0is0kikfw7h9w3n8ikw70bargr32d1cyasqgmb7s7mvs1c
got: sha256:1qs5p05qfk5xs1ajwyhn27m0bzs96lnlf3b4gnkffajhaq7hiqll
cannot build derivation '/nix/store/aj5ag721b9gm4an6yxh2ljg19ixg4alv-dhall-nix-simple.drv': 1 dependencies couldn't be built
The reason why this happens is because GitHub's tarballs are not
deterministic and whenever GitHub changes something in the way these are
generated, we get a hash mismatch.
For exactly that reason, the fetchFromGitHub wrapper in nixpkgs uses
fetchzip instead of fetchurl, so that file ordering in the archive
doesn't matter.
Unfortunately, the upstream project still uses fetchurl, but since the
URLs and hashes have changed due to a bump to Dhall version 1.33.1, I've
choosen to switch to latest master instead of monkeypatching via
extraPostFetch.
With this bandaid, we shouldn't run into a hash collision until either
the next GC or until the upstream project has switched to either
fetchFromGitHub or fetchzip.
Signed-off-by: aszlig <aszlig@nix.build>
Cc: @Profpatsch, @justinwoo
|
|
|
|
| |
Stuff like `virt-manager` has broken buttons.
|
| |
|
| |
|
|
|
|
|
|
|
|
|
| |
I want to be able to open http(s) links that are e.g. images directly
in the right application. Aka web urls should be transparent, instead
of always opening everthing in the browser.
This adds some silly ways of connecting to the server and parsing
out the headers, in order to fetch the content-type.
|
|
|
|
|
|
|
|
| |
This is an experiment about whether we can get away with using the
non-recursive version by default.
The U::Record variant uses a Vec instead of a HashMap by default, to
make encoding from lists easier, and keep the ordering as given.
|
| |
|
|
|
|
|
|
|
|
| |
It’s a lot simpler to just export the parsed attribute as envvars.
Remove the substitute stuff (it already went into the el_substitute
lib anyway) and replace the xpathexec0 code with the function from the
el_exec lib.
|
|
|
|
|
| |
A small parser for http/https URLs.
Substitutes host/port/path in argv.
|
| |
|
|
|
|
| |
https://github.com/NixOS/nixpkgs/commit/2c616aa63895f064ac1b1332a8cf9e7ae3ddbb44
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
| |
This was previously located verbatim on my webserver.
Since `df.eu` thought it was a good idea to unilaterally cancel it
when I moved my domain, it is now a good idea to nixify what was
there.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The C implementation of el_semicolon in execline only reads one
argument at a time and returns an index into the rest of argv.
This makes sense for the usual application of programs using it, which
is just reading a few arguments and a block or two, and then executing
into `prog`. `prog` could be anything really, including additional
blocks.
The new `el_semicolon_full_argv` function exports the previous
behaviour of parsing the whole thing.
As a nice side-effect, we return the rest of argv in-place.
|
|
|
|
|
|
|
|
|
|
|
| |
el_exec: wraps the various execve wrappers in skalib that are useful
for writing execline-like utils. currently only `xpathexec0` is
supported, which execs into the argv you give it or errors with the
right error if file not found.
el_substitute: execline argv substitution! Wraps the execline
function, so it will behave exactly the same as the existing execline
utils, like `importas`.
|
| |
|
|
|
|
|
| |
We can define a more or less complete generator in less than 50 lines
of nix. Nice.
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
A while ago[1], @Profpatsch added this comment above the definition of
the $runtimeDeps variable:
# Reads the dependency closures and does ? something? TODO: explain
I just recently (yesterday as of the date of this commit) found out
about that comment by accident.
While this should probably be better of as an issue instead, the comment
does have a point, since not everybody enjoys reading/writing sed
expressions.
In a nutshell, what the implementation actually does is parsing the
output of the files generated by exportReferencesGraph.
At the time of writing the implementation, we didn't have a JSON-based
interface in Nix for doing the same, nor did we have something like
pkgs.closureInfo. There was only a small Perl script[2], which did
something like this, but given that it can be easily done via sed, I
opted to instead use the latter.
Nevertheless however, using closureInfo is not only more concise in its
implementation, it also makes our implementation much more concise as
well and also obvious on what we're doing here.
[1]: 09dc1d8ad625b9a1d5b89593b184d316837ba1cc
[2]: https://github.com/NixOS/nixpkgs/blob/8747190024205a5a3534b4e9a18dbaf3f3ee7b39/pkgs/build-support/kernel/paths-from-graph.pl
Signed-off-by: aszlig <aszlig@nix.build>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Most of the games are using "en3installer0" as the downloadName, so it
makes sense to use this as the default, since first of all do not
support languages other than English except for a few games (eg. Albion)
and second, there is no game with several installer packages to chose
from (at least not on top of my head).
Even if there would be, it would still be the exception rather than the
rule, so let's optimise for the common scenario.
Signed-off-by: aszlig <aszlig@nix.build>
|
|
|
|
|
|
|
|
| |
I've had this in my configuration.nix for a while and since I got used
to it, I think it's a good idea to enable both Nix Flakes and the "nix"
command by default for all my machines.
Signed-off-by: aszlig <aszlig@nix.build>
|
|
|
|
|
| |
Instead of adding a new type, it just uses the 2^1 natural, which has
exactly two possibilities.
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
| |
Also add the service to legosi so I can use it from the weechat user.
|
| |
|
|
|
|
| |
No libpurple for just XMPP, phew.
|
|
|
|
| |
I'm fairly certain this isn't needed anymore.
|
|
|
|
|
|
|
| |
According to [1], `udev` has gained native support for FIDO security
tokens, and subsequently has been removed upstream.
[1]: 89c3e73dad0970b26183e415555fb0379ba33e7a
|
|
|
|
|
|
|
| |
As already stated otherwise in this log, it's back, so reverting the
removal.
This reverts commit 54b3f8ac7da02738be6919d05ae12d5c14796e3f.
|
|
|
|
|
|
|
|
|
|
|
|
| |
This reverts commit 64fdbcc7e0949a3d3bf1cee3361d5ebe5c56cc4c.
It seems that there is now[1] a maintained version of p7zip and since I
didn't get used to unar so far, I take this opportunity to switch back
to p7zip.
[1]: https://github.com/NixOS/nixpkgs/pull/90140
Signed-off-by: aszlig <aszlig@nix.build>
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
While it's fine to use <nixpkgs> on most systems, we deliberately want
to avoid the use of <nixpkgs> to make sure that whenever we for example
run <nixpkgs/nixos/lib/eval-config.nix> with a custom "pkgs" argument we
are guaranteed that we get the version we specify.
So this is one of the reason I used <nixpkgsSrc> on Hydra instead of
<nixpkgs>, so that whenever we have such occasions where we can't
guarantee such things, the evaluation will fail.
And right now, it does:
in job 'machines.profpatsch.legosi':
file 'nixpkgs/nixos/modules/profiles/qemu-guest.nix' was not found in
the Nix search path (add it using $NIX_PATH or -I), at
.../machines/profpatsch/legosi.nix:12:5
Fortunately, there is modulesPath, which refers to
<nixpkgs/nixos/modules> of the nixpkgs version passed via "pkgs".
Signed-off-by: aszlig <aszlig@nix.build>
Cc: @Profpatsch
|