This update fixes an issue with the server not rendering the webui, see: https://github.com/ggml-org/llama.cpp/pull/20724
* gnu/packages/machine-learning.scm (llama-cpp): Update to 0.0.0-b8445.
Change-Id: Ib90076dfa8af6983587bf8408a465c34b4872873
This commit was made by running this command:
sed -e's/modify-inputs (package-\([a-z-]*\)inputs [a-zA-Z0-9-]\+)/modify-inputs \1inputs/g' -i gnu/packages/*.scm
… and then reverting individual hunks where the change would trigger unbound
variable warnings or other issues (such as ‘native-inputs’ is bound in the
body of the ‘inputs’ field, but it refers to the ‘native-inputs’ thunk defined
just above).
Change-Id: I6d94819f2809313fa1fbefc61897502ee7d66fab
* gnu/packages/astronomy.scm (python-astropy-minimal)[native-inputs]: Add
‘specification-astropy-data’, referred to via ‘this-package-input’ in the
inherited phases.
Change-Id: I8d287176f9807061dc6dde453dacc9a15f209746
* guix/build-system/gnu.scm (package-with-explicit-inputs/deprecated): In
the field value bodies of inherited packages, use the field name to refer to
the inherited value.
(package-with-explicit-inputs*)[add-explicit-inputs]: Likewise.
(package-with-extra-configure-variable): Likewise.
(static-package): Likewise.
(dist-package): Likewise.
(package-with-restricted-references): Likewise.
Change-Id: I62a95563cbdd48170a674b4995585c8778808b8c
This commit was made by running this command:
sed -e's/substitute-keyword-arguments (package-arguments [a-zA-Z0-9-]\+)/substitute-keyword-arguments arguments/g' -i gnu/packages/*.scm
… and then:
1. reverting changes from ‘gnu/packages/rust.scm’ and
‘gnu/packages/java.scm’ since they would incur derivation changes and/or
breakage;
2. reverting the change for ‘gcc-final’ in ‘gnu/packages/commencement.scm’;
3. reverting the change for ‘onnx-optimizer’, ‘openquest’, and ‘certbot’,
which use ‘substitute-keyword-arguments’ for arguments that are not
inherited (and thus ‘arguments’ would be unbound);
4. reverting the change for ‘insight-toolkit-legacy’ and ‘wine64-staging’
which make bogus assumptions about inherited arguments.
Change-Id: I122a7cf517b6b63cae38944b5d33ade4b1f5a89c
* guix/records.scm (make-syntactic-constructor)[field-index]: New procedure.
[wrap-field-value]: Add optional argument ‘parent’. When it is true, bind F
to the inherited field value.
[field-bindings/inheritance]: New procedure.
Use it.
* tests/records.scm ("define-record-type* & thunked & no inherited value")
("define-record-type* & thunked & inherited value")
("define-record-type* & thunked & inherited value & this-record"): New tests.
* doc/guix.texi (Defining Package Variants): Update ‘modify-inputs’ example to
refer to ‘inputs’.
(Writing Manifests): Likewise.
* doc/guix-cookbook.texi (Package Variants): Likewise for
‘substitute-keyword-arguments’.
Fixes: https://issues.guix.gnu.org/50335
Change-Id: If4e18155ce203637ff9e116ee8098f8997bfebe2
With the next commit, ‘inputs’ is bound within the body of the ‘inputs’ field,
which would shadow the ‘inputs’ parameter of ‘package-with-explicit-inputs*’.
* guix/build-system/gnu.scm (package-with-explicit-inputs*): Rename ‘inputs’
parameter to ‘explicit-inputs’ to avoid name class in ‘add-explicit-inputs’.
Change-Id: I5aa70504cc6ffc856728c6e1a715ef68d9ebd7d9
Test groups were not used in any meaningful way. The group path was not
printed and it was not used in test selection mechanism. I think groups are
useful, and it is nice to be able to, for example, run tests from a single
group.
This commit does two things. First, it changes the test reporting to include
the value returned from test-runner-group-path, so you will know not only the
test name, but the test group(s) as well. And second, it changes the test
selection (and exclusion) process to match against the "full" test name, so
group path + test name.
Hence
(test-begin "failing tests")
(test-equal "this should fail" 1 2)
(test-end)
will, depending on the output location, produce following text.
.trs:
:test-result: FAIL failing tests: this should fail [0.000s]
:test-global-result: FAIL
:recheck: yes
:copy-in-global-log: yes
.log:
test-name: failing tests: this should fail
location: test.scm:140
source:
+ (test-equal "this should fail" 1 2)
expected-value: 1
actual-value: 2
result: FAIL
stdout:
FAIL: test.scm - failing tests: this should fail [0.000s]
* build-aux/test-driver.scm (current-test-full-name): New procedure.
(test-runner-gnu): Use current-test-full-name instead of
test-runner-test-name.
(test-match-name*): Match against current-test-full-name. Use compose.
(test-match-name*/negated): Rewrite in terms of test-match-name*.
Change-Id: I3fb9a2a721165204f020b79e019533f799b790e4
Signed-off-by: Maxim Cournoyer <maxim@guixotic.coop>
Modified-by: Maxim Cournoyer <maxim@guixotic.coop>
The :test-global-result: .trs metadata contained just either FAIL, SKIP or
PASS with a comment that further refinements are required for XPASS. The
description of :test-global-result: is in the manual is as follows:
This is used to declare the "global result" of the script.
Currently, the value of this field is needed only to be reported
(more or less verbatim) in the generated global log file
‘$(TEST_SUITE_LOG)’, so it’s quite free-form. For example, a test
script which runs 10 test cases, 6 of which pass and 4 of which are
skipped, could reasonably have a ‘PASS/SKIP’ value for this field,
while a test script which runs 19 successful tests and one failed
test could have an ‘ALMOST PASSED’ value.
As we can see, the examples as `PASS/SKIP' and `ALMOST PASSED', so there is no
need to stick to strict model. Hence this commit changes the resulting value
to be comma-separated list of PASS, FAIL, XPASS, XFAIL and SKIP. The
respective elements are present only when the count of tests with such a
result is positive.
In practice, that should usually produce lines such as
:test-global-result: PASS,FAIL
or
:test-global-result: PASS
* build-aux/test-driver.scm (test-runner-gnu)[finalize]: Refine the output of
:test-global-result:.
Change-Id: I7178ac9703e1749adf6de2445f7ed0591983cef2
Signed-off-by: Maxim Cournoyer <maxim@guixotic.coop>
* gnu/packages/patches/lufa-fix-incompatible-cast.patch: New file.
* gnu/local.mk (dist_patch_DATA): Register it.
* gnu/packages/avr-xyz.scm (lufa)[source]: Use it.
Followup to c582fb3124 which incorrectly used
the utf8-locales.
* gnu/packages/gnome.scm (geocode-glib)[native-inputs]: Replace
libc-utf8-locales-for-target by libc-locales-for-target.
[arguments]<#:phases>: Remove 'set-locales.
Fixes: guix/guix#7304
Change-Id: If8c17d788cda767807bad168076e37eef932b64c
Signed-off-by: Andreas Enge <andreas@enge.fr>
Current implementation printed metadata supposed to be present just once per
.trs file on the end of each test group. According to the automake's manual
that is undefined behavior. This commit fixes it by printing that metadata
just once, after all tests did run.
Since there is no built-in hook that could be used for
that (test-runner-on-final runs on *each* outermost test-end), I introduced
new `finalize' procedure that need to be called by the user. Possibly not the
most elegant solution, but since we are the only user, it works fine and
produces actually valid .trs file.
That also means there is no longer any use for test-runner-on-test-end!.
* build-aux/test-driver.scm (test-runner-gnu): Define new procedure `finalize'
and return it together with the runner. Do not call
test-runner-on-group-end!.
(main): Call the `finalize' after all tests are done.
Signed-off-by: Maxim Cournoyer <maxim@guixotic.coop>
Guile 3.0.11 ships with a new SRFI 64.
* build-aux/test-driver.scm (%test-match-all): Delete procedure.
(main): Replace %test-match-all with test-match-all.
Change-Id: Ibcd7d98194694c1de679491be4f5a603a79f6e29
* nix/libstore/build.cc (prepareSlirpChrootAction): Use
canonPath(i, true) to resolve symlinks when adding /etc/resolv.conf
and /etc/hosts to the slirp4netns chroot, so that bindMount receives
a regular file path instead of a symlink.
On systems using systemd-resolved, /etc/resolv.conf is typically a
symlink:
/etc/resolv.conf -> /run/systemd/resolve/stub-resolv.conf
The slirp4netns chroot creates an empty /run/ directory, so when
bindMount copies the symlink verbatim (spawn.cc line 537-542), the
target does not exist and slirp4netns cannot determine the upstream
DNS server. This causes all DNS resolution to fail for fixed-output
derivations that use the Guile-based git-fetch builder (e.g.
git-fetch/lfs), since they rely on slirp4netns for network access
in the build chroot.
Derivations using builtin:git-download are unaffected because they
run in the daemon process itself, which has full network access.
Change-Id: Ib73e69a8760e74eb8141dd0408c27aa8b3001e37
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
Merges: #6959
