See http://debbugs.gnu.org/cgi/bugreport.cgi?bug=37318.
* gnu/services/networking.scm (openntpd-configuration->string): New procedure,
extracted from top of the `openntpd-shepherd-service' to make it testable.
(openntpd-shepherd-service): Adapt following the move of the code to the above
procedure.
* tests/networking.scm: Add a test for the `openntpd-configuration->string'
procedure.
* gnu/services/networking.scm (ntp-server-types): New enum.
(<ntp-server>): New record type.
(ntp-server->string): New procedure.
(%ntp-servers): Define in terms of <htp-server> records. Use the first
entrypoint server as a pool instead of a list of static servers. This is more
resilient since a new server of the pool can be interrogated on every
request. Add the 'iburst' options.
(ntp-configuration-servers): Define a custom accessor that warns but honors
the now deprecated server format.
(<ntp-configuration>): Use it.
(%openntpd-servers): New variable,
(<openntpd-configuration>): Use it, as a pool ('servers' field) instead of a
regular server.
* tests/networking.scm: New file.
* Makefile.am (SCM_TESTS): Register it.
* doc/guix.texi: Update documentation.
* guix/import/crate.scm (%dual-license-rx): New variable.
(crate->guix-package)[string->license]: Rewrite to match it.
* tests/crate.scm (test-crate): Adjust "license" field to current
practice.
This provides a clean separation between bindings to the
https://crates.io/api/v1 API and actual conversion to Guix package
sexps.
As a side-effect, it fixes things like "guix import blake2-rfc", "guix
refresh -t crates", etc.
* guix/import/crate.scm (<crate>, <crate-version>, <crate-dependency>):
New record types.
(lookup-crate, crate-version-dependencies): New procedures.
(crate-fetch): Remove.
(crate->guix-package): Rewrite to use the new API.
(latest-release): Likewise.
* guix/build-system/cargo.scm (%crate-base-url): New variable.
* tests/crate.scm (test-crate): Update accordingly.
fixlet
* guix/swh.scm (%allow-request?, %save-rate-limit-reset-time)
(%general-rate-limit-reset-time): New variables.
(request-rate-limit-reached?, update-rate-limit-reset-time!): New
procedures.
(call): Call '%allow-request?'. Change 'swh-error' protocol to pass
METHOD in addition to URL.
* tests/swh.scm ("rate limit reached")
("%allow-request? and request-rate-limit-reached?"): New tests.
* guix/tests/http.scm (call-with-http-server): Replace 'code' and 'data'
parameters with 'responses+data'. Compute RESPONSES as a function of
that. Remove #:headers parameter.
[http-write]: Quit only when RESPONSES is empty.
[server-body]: Get the response and data from RESPONSES, and set it to
point to the rest.
(with-http-server): Adjust accordingly.
* tests/derivations.scm ("'download' built-in builder")
("'download' built-in builder, invalid hash")
("'download' built-in builder, not found")
("'download' built-in builder, check mode"): Adjust to new
'with-http-server' interface.
* tests/lint.scm ("home-page: 200")
("home-page: 200 but short length")
("home-page: 404", "home-page: 301, invalid"):
("home-page: 301 -> 200", "home-page: 301 -> 404")
("source: 200", "source: 200 but short length")
("source: 404", "source: 404 and 200")
("source: 301 -> 200", "source: 301 -> 404"):
("github-url", github-url): Likewise.
* tests/swh.scm (with-json-result)
("lookup-origin, not found"): Likewise.
When adding multiple instances of a service requiring some user
account/group, we could end up with multiple entries for that account or
group in /etc/passwd or /etc/group.
* gnu/build/accounts.scm (database-writer)[write-entries]: Add call to
'delete-duplicates'.
* tests/accounts.scm ("write-passwd with duplicate entry"): New test.
Fixes <https://bugs.gnu.org/37161>.
* guix/scripts/pack.scm (docker-image)[build]: Add a 'directory' entry
for "/tmp" to DIRECTIVES.
* tests/pack.scm ("docker-image + localstatedir"): Test the presence of /tmp.
* gnu/tests/docker.scm (run-docker-test)["Load docker image and run
it"]: Test the presence and permission bits of "/tmp".
Fixes <https://bugs.gnu.org/36925>.
Reported by Jesse Gibbons <jgibbons2357@gmail.com>.
* guix/scripts/pack.scm (wrapped-package): Add 'output*' parameter.
[build]: Define 'input' and 'target'; use them instead of #$package and
#$output, respectively.
(wrapped-manifest-entry): New procedure.
(map-manifest-entries): Call PROC directly.
(guix-pack): Pass WRAPPED-MANIFEST-ENTRY to 'map-manifest-entries'.
In guile-json version 3, JSON objects are represented as hash tables, rather
than alists.
* guix/import/cpan.scm (string->license): Change the match expression to match
on lists, rather than vectors.
(module->dist-name, cpan-source-url, cpan-version): Change assoc-ref to
hash-ref.
(cpan-module->sexp): Change assoc-ref to hash-ref, and assoc-ref* to
hash-ref*.
* tests/cpan.scm ("source-url-http", "source-url-https"): Convert the alist to
a hash table.
This ensures that this test is skipped in contexts where user namespaces
are not supported, instead of failing.
This is a followup to commit b6dc08393e.
* tests/guix-environment.sh: Move '--container --no-cwd' test to...
* tests/guix-environment-container.sh: ... here.
Fixes a regression introduced in
50fc2384fe.
* tests/lint.scm (warning-contains?): New procedure.
("home-page: host not found"): Use 'warning-contains?' instead of
testing for equality, as was the case before commit 50fc2384fe.
This handles the case where the 'getaddrinfo' error is not "Name or
service not known" but instead something like "System error" or
"Servname not supported for ai_socktype", as is the case in the build
environment.
Fixes <https://bugs.gnu.org/36777>.
Reported by Carl Dong <contact@carldong.me>.
* guix/derivations.scm (derivation/masked-inputs): Call
'delete-duplicates' on INPUTS.
* tests/derivations.scm ("derivation with duplicate fixed-output inputs"):
New test.
Fixes <https://bugs.gnu.org/36813>.
Reported by Jakob L. Kreuze <zerodaysfordays.sdf.org@sdf.org>.
* guix/gexp.scm (program-file-compiler): Pass #:system and #:target to
'gexp->script'.
(load-path-expression): Add #:system and #:target and honor them.
(gexp->script): Likewise.
* tests/gexp.scm ("program-file #:system"): New test.
* doc/guix.texi (G-Expressions): Adjust accordingly.
Guile-JSON 3.x is incompatible with Guile-JSON 1.x, which we relied on
until now: it maps JSON dictionaries to alists (instead of hash tables),
and JSON arrays to vectors (instead of lists). This commit is about
adjusting all the existing code to this new mapping.
* m4/guix.m4 (GUIX_CHECK_GUILE_JSON): New macro.
* configure.ac: Use it.
* doc/guix.texi (Requirements): Mention the Guile-JSON version.
* guix/git-download.scm (git-fetch)[guile-json]: Use GUILE-JSON-3.
* guix/import/cpan.scm (string->license): Expect vectors instead of
lists.
(module->dist-name): Use 'json-fetch' instead of 'json-fetch-alist'.
(cpan-fetch): Likewise.
* guix/import/crate.scm (crate-fetch): Likewise, and call 'vector->list'
for DEPS.
* guix/import/gem.scm (rubygems-fetch): Likewise.
* guix/import/json.scm (json-fetch-alist): Remove.
* guix/import/pypi.scm (pypi-fetch): Use 'json-fetch' instead of
'json-fetch-alist'.
(latest-source-release, latest-wheel-release): Call 'vector->list' on
RELEASES.
* guix/import/stackage.scm (stackage-lts-info-fetch): Use 'json-fetch'
instead of 'json-fetch-alist'.
(lts-package-version): Use 'vector->list'.
* guix/import/utils.scm (hash-table->alist): Remove.
(alist->package): Pass 'vector->list' on the inputs fields, and default
to the empty vector.
* guix/scripts/import/json.scm (guix-import-json): Remove call to
'hash-table->alist'.
* guix/swh.scm (define-json-reader): Expect pair? or null? instead of
hash-table?.
[extract-field]: Use 'assoc-ref' instead of 'hash-ref'.
(json->branches): Use 'map' instead of 'hash-map->list'.
(json->checksums): Likewise.
(json->directory-entries, origin-visits): Call 'vector->list' on the
result of 'json->scm'.
* tests/import-utils.scm ("alist->package with dependencies"): New test.
* gnu/installer.scm (build-compiled-file)[builder]: Use GUILE-JSON-3.
* gnu/installer.scm (installer-program)[installer-builder]: Likewise.
* gnu/installer/locale.scm (iso639->iso639-languages): Use 'assoc-ref'
instead of 'hash-ref', and pass vectors through 'vector->list'.
(iso3166->iso3166-territories): Likewise.
* gnu/system/vm.scm (system-docker-image)[build]: Use GUILE-JSON-3.
* guix/docker.scm (manifest, config): Adjust for Guile-JSON 3.
* guix/scripts/pack.scm (docker-image)[build]: Use GUILE-JSON-3.
* guix/import/github.scm (fetch-releases-or-tags): Update docstring.
(latest-released-version): Use 'assoc-ref' instead of 'hash-ref'. Pass
the result of 'fetch-releases-or-tags' to 'vector->list'.
* guix/import/launchpad.scm (latest-released-version): Likewise.
This is a followup to f363c836e0.
* tests/lint.scm ("cve")
("cve: one vulnerability")
("cve: one patched vulnerability")
("cve: known safe from vulnerability")
("cve: vulnerability fixed in replacement version")
("cve: patched vulnerability in replacement"): Refer to
'package-vulnerabilities' from (guix lint), not (guix scripts lint).
* guix/discovery.scm (scheme-modules): Rename the inner 'file' to
'relative'. Pass FILE as an addition argument to WARN.
* guix/ui.scm (warn-about-load-error): Add 'module' argument (actually,
what was called 'file' really contained a module name.) Call
'check-module-matches-file' in the catch-all error case.
(check-module-matches-file): New procedure.
* tests/guix-build.sh: Test it.
This simplifies the code since one no longer needs to think about
whether '.guix-channel' was present.
* guix/channels.scm (read-channel-metadata): Always pass a string as the
first argument to 'channel-metadata'.
(read-channel-metadata-from-source): Always return a <channel-metadata>
record.
(channel-instance-dependencies): Remove now unneeded 'match'.
(standard-module-derivation): Assume DIRECTORY is never #f and contains
a leading slash.
* tests/channels.scm (channel-metadata-directory)
(channel-metadata-dependencies): New procedures.
("channel-instance-metadata returns #f if .guix-channel does not
exist"): Remove.
("channel-instance-metadata returns default if .guix-channel does not
exist"): New test.
(make-instance): Use 'write' instead of 'display' when creating
'.guix-channel'.
(instance--no-deps): Remove dependencies.
(instance--sub-directory): New variable.
("channel-instance-metadata and default dependencies")
("channel-instance-metadata and directory"): New tests.
("latest-channel-instances excludes duplicate channel dependencies"):
Expect 'channel-commit' to return a string and adjust accordingly.
Until now the 'version' field in '.guix-channel' could be omitted, or it
could be any value.
* guix/channels.scm (read-channel-metadata): Rename to...
(channel-instance-metadata): ... this.
(channel-instance-dependencies): Adjust accordingly.
(read-channel-metadata): New procedure. Use 'match'
to require a 'version' field. Provide proper error handling when the
channel sexp is malformed or when given an unsupported version number.
(read-channel-metadata-from-source): Use 'catch' and
'system-error-errno' instead of 'file-exists?'.
* tests/channels.scm (instance--unsupported-version): New variable.
(read-channel-metadata): Rename to...
(channel-instance-metadata): ... this. Rename tests accordingly.
("channel-instance-metadata rejects unsupported version"): New test.
Rather than emiting warnings directly to a port, have the checkers return the
warning or warnings.
This makes it easier to use the warnings in different ways, for example,
loading the data in to a database, as you can work with the <lint-warning>
records directly, rather than having to parse the output to determine the
package and location.
* guix/scripts/lint.scm (<lint-warning>): New record type.
(lint-warning): New macro.
(lint-warning?, lint-warning-package, lint-warning-message,
lint-warning-location, package-file, make-warning): New procedures.
(call-with-accumulated-warnings, with-accumulated-warnings): Remove.
(emit-warning): Rename to emit-warnings, and switch to displaying multiple
warnings.
(check-description-style)[check-not-empty-description, check-texinfo-markup,
check-trademarks, check-quotes, check-proper-start,
check-end-of-sentence-space]: Switch to generating a list of warnings, and
using make-warning, rather than emit-warning.
(check-inputs-should-be-native, check-inputs-should-not-be-an-input-at-all):
Switch to generating a list of warnings, and using make-warning, rather than
emit-warning.
(check-synopsis): Switch to generating a list of warnings, and using
make-warning, rather than emit-warning.
[check-not-empty]: Remove, this is handled in the match clause
to avoid other warnings being emitted.
[check-final-period, check-start-article, check-synopsis-length,
check-proper-start, check-start-with-package-name, check-texinfo-markup]:
Switch to generating a list of warnings, and using make-warning, rather than
emit-warning.
[checks]: Remove check-not-empty.
(validate-uri, check-home-page, check-patch-file-names,
check-gnu-synopsis+description): Switch to generating a list of warnings, and
using make-warning, rather than emit-warning.
(check-source): Switch to generating a list of warnings, and using
make-warning, rather than emit-warning.
[try-uris]: Remove.
[warnings-for-uris]: New procedure, replacing try-uris.
(check-source-file-name, check-source-unstable-tarball, check-mirror-url,
check-github-url, check-derivation, check-vulnerabilities, check-for-updates,
report-tabulations, report-trailing-white-space, report-long-line,
report-lone-parentheses, report-formatting-issues, check-formatting): Switch
to generating a list of warnings, and using make-warning, rather than
emit-warning.
(run-checkers): Call emit-warnings on the warnings returned from the checker.
* tests/lint.scm (string-match-or-error, single-lint-warning-message): New
procedures.
(call-with-warnings, with-warnings): Remove.
("description: not a string", "description: not empty", "description: invalid
Texinfo markup", "description: does not start with an upper-case letter",
"description: may start with a digit", "description: may start with lower-case
package name", "description: two spaces after end of sentence", "description:
end-of-sentence detection with abbreviations", "description: may not contain
trademark signs: ™", "description: may not contain trademark signs: ®",
"description: suggest ornament instead of quotes", "synopsis: not a string",
"synopsis: not empty", "synopsis: valid Texinfo markup", "synopsis: does not
start with an upper-case letter", "synopsis: may start with a digit",
"synopsis: ends with a period", "synopsis: ends with 'etc.'", "synopsis:
starts with 'A'", "synopsis: starts with 'a'", "synopsis: starts with 'an'",
"synopsis: too long", "synopsis: start with package name", "synopsis: start
with package name prefix", "synopsis: start with abbreviation", "inputs:
pkg-config is probably a native input", "inputs: glib:bin is probably a native
input", "inputs: python-setuptools should not be an input at all (input)",
"inputs: python-setuptools should not be an input at all (native-input)",
"inputs: python-setuptools should not be an input at all (propagated-input)",
"patches: file names", "patches: file name too long", "patches: not found",
"derivation: invalid arguments", "license: invalid license", "home-page: wrong
home-page", "home-page: invalid URI", "home-page: host not found", "home-page:
Connection refused", "home-page: 200", "home-page: 200 but short length",
"home-page: 404", "home-page: 301, invalid", "home-page: 301 -> 200",
"home-page: 301 -> 404", "source-file-name", "source-file-name: v prefix",
"source-file-name: bad checkout", "source-file-name: good checkout",
"source-file-name: valid", "source-unstable-tarball",
"source-unstable-tarball: source #f", "source-unstable-tarball: valid",
"source-unstable-tarball: package named archive", "source-unstable-tarball:
not-github", "source-unstable-tarball: git-fetch", "source: 200", "source: 200
but short length", "source: 404", "source: 301 -> 200", "source: 301 -> 404",
"mirror-url", "mirror-url: one suggestion", "github-url", "github-url: one
suggestion", "github-url: already the correct github url", "cve", "cve: one
vulnerability", "cve: one patched vulnerability", "cve: known safe from
vulnerability", "cve: vulnerability fixed in replacement version", "cve:
patched vulnerability in replacement", "formatting: lonely parentheses",
"formatting: alright"): Change test-assert to test-equal, and adjust to work
with the changes above.
("formatting: tabulation", "formatting: trailing white space", "formatting:
long line"): Use string-match-or-error rather than string-contains.
* guix/gexp.scm (lower-inputs): Return either <derivation-input> records
or store items.
(lower-reference-graphs): Return file/input pairs.
(<lowered-gexp>)[sources]: New field.
(lower-gexp): Adjust accordingly.
(gexp->input-tuple): Remove.
(gexp->derivation)[graphs-file-names]: Handle only the
'derivation-input?' and 'string?' cases.
Pass #:sources to 'raw-derivation'; ensure #:inputs contains only
<derivation-input> records.
* guix/remote.scm (remote-eval): Adjust to the new <lowered-gexp>
interface.
* tests/gexp.scm ("lower-gexp"): Adjust to expect <derivation-input>
records instead of <gexp-input>
* doc/guix.texi (Invoking guix environment): Add --no-cwd.
* guix/scripts/environment.scm (show-help, %options): Add --no-cwd.
(launch-environment/container): Add 'map-cwd?' param; only add mapping
for cwd if #t. Only change to cwd within container if #t, otherwise
home.
(guix-environment): Error if --no-cwd without --container. Provide
'(not no-cwd?)' to launch-environment/container as 'map-cwd?'.
* tests/guix-environment.sh: Add test for no-cwd.
Co-authored-by: Mike Gerwitz <mtg@gnu.org>
This fixes a bug whereby "guix build texlive -n" would report:
0.0 MB would be downloaded:
/gnu/store/…-texlive-20180414
instead of:
The following derivation would be built:
/gnu/store/…-texlive-texmf-20180414.drv
2,595.2 MB would be downloaded:
/gnu/store/…-texlive-20180414-texmf.tar.xz
/gnu/store/…-texlive-20180414
where 'texlive-texmf' is a non-substitutable dependency of 'texlive'.
* guix/derivations.scm (dependencies-of-substitutables): New procedure.
(derivation-build-plan): When 'input-substitutable-info' returns true,
append the subset of DEPS that corresponds to SUBSTITUABLES to the first
argument of 'loop'.
* guix/ui.scm (show-what-to-build): Remove half-baked traversal of
DOWNLOAD.
* tests/derivations.scm ("derivation-build-plan and substitutes,
non-substitutable dep"): New test.
* guix/gexp.scm (gexp-input-thing, gexp-input-output)
(gexp-input-native?): Export.
(lower-inputs): Return <gexp-input> records instead of tuples.
(lower-reference-graphs): Adjust accordingly.
(<lowered-gexp>): New record type.
(lower-gexp, gexp-input->tuple): New procedure.
(gexp->derivation)[%modules]: Remove.
[requested-graft?]: New variable.
[add-modules]: New procedure.
Rewrite in terms of 'lower-gexp'.
(gexp-inputs): Add TODO comment.
* tests/gexp.scm ("lower-gexp"): New test.
* guix/import/pypi.scm (maybe-inputs): Add INPUT-TYPE argument, and use it.
(test-section?): New predicate.
(parse-requires.txt): Collect the optional test inputs, and return them as the
second element of the returned list.
(parse-wheel-metadata): Likewise.
(guess-requirements): Adapt.
(make-pypi-sexp): Likewise, and include the test inputs requirements as native
inputs in the returned package expression.
* tests/pypi.scm (test-requires.txt): Include a test section in the
test-requires.txt data.
(test-requires.txt-beaker): New variable.
("parse-requires.txt"): Adapt.
("parse-requires.txt - Beaker"): New test.
("parse-wheel-metadata, with extras"): Adapt.
("parse-wheel-metadata, with extras - Jedi"): Adapt.
("pypi->guix-package, no wheel"): Re-indent, and add the expected
native-inputs.
("pypi->guix-package, wheels"): Likewise.
With newer Wheel releases, there is no more metadata.json file; the METADATA
file should be used instead (see: https://github.com/pypa/wheel/issues/195).
This change updates our PyPI importer so that it uses the latter.
* guix/import/pypi.scm (define-module): Remove unnecessary modules and export
the PARSE-WHEEL-METADATA procedure.
(parse-wheel-metadata): Add procedure.
(guess-requirements): Use it.
* tests/pypi.scm (test-metadata): Test it.
This change enables the PyPI importer to look for requirements in a source
archive of a different type than "tar.gz" or "tar.bz2". Also, scan the source
archive to find a requires.txt file.
* guix/import/pypi.scm: (guess-requirements)[tarball-directory]: Remove procedure.
[guess-requirements-from-source]: Use COMRESSED-FILE? to determine if an
archive type is supported, and some file extension logic that chooses either
"tar" or "unzip" as the extractor. Search for the requires.txt file in the
archive instead of using a static, expected location.
(guess-requirements): Rename the TARBALL argument to ARCHIVE, to denote the
archive format is no longer bound specifically to the Tar format.
(compute-inputs): Likewise.
* tests/pypi.scm ("pypi->guix-package, no wheel"): Mock the requires.txt at a
non-standard location.
("pypi->guix-package, no usable requirement file."): New test.
The previous solution was fragile and could leave unwanted characters in a
requirement name, such as '[' or ']'.
Partially fixes <https://bugs.gnu.org/33047>.
* guix/import/pypi.scm (use-modules): Export SPECIFICATION->REQUIREMENT-NAME
(%requirement-name-regexp): New variable.
(clean-requirement): Rename to...
(specification->requirement-name): this, which now uses
%requirement-name-regexp to select the requirement name from the requirement
specification.
(parse-requires.txt): Adapt.
* guix/import/pypi.scm: Export PARSE-REQUIRES.TXT.
(clean-requirement): Move procedure to the top level.
(guess-requirements): Move the READ-REQUIREMENTS procedure to the top level,
and rename it to PARSE-REQUIRES.TXT. Move the CLEAN-REQUIREMENT procedure to
the top level. Move the COMMENT? functions inside the PARSE-REQUIRES.TXT
procedure.
(parse-requires.txt): Add a SECTION-HEADER? predicate, and use it to prevent
parsing optional requirements.
* tests/pypi.scm (test-requires-with-sections): New variable.
("parse-requires.txt, with sections"): New test.
The output of tar when creating archives for the purpose of tests is not
useful, so we mute it.
* tests/pypi.scm ("pypi->guix-package"): Mute the output of tar.
("pypi->guix-package, wheels"): Likewise.
PyPI packages are mandated to have a setup.py file, which contains a listing
of the required dependencies. The setuptools/distutils machinery embed
metadata in the archives they produce, which contains this information. There
is no need nor gain to collect the requirements from a "requirements.txt"
file, as it is not the true record of dependencies for PyPI packages and may
contain extraneous requirements or not exist at all.
* guix/import/pypi.scm (guess-requirements): Update comment.
[guess-requirements-from-source]: Do not attempt to parse the file
requirements.txt. Streamline logic.
* tests/pypi.scm (test-requires.txt): Rename from test-requirements, to hint
at the file being tested.
("pypi->guix-package"): Adapt so that the fake package contains a requires.txt
file rather than a requirements.txt file.
("pypi->guix-package, wheels"): Likewise.
* guix/import/crate.scm (maybe-cargo-inputs): Remove one level of lists.
* guix/import/crate.scm (maybe-cargo-development-inputs): Same.
* tests/crate.scm: (crate->guix-package)[package]<#:arguments>: Remove
one level of list nesting.
This could lead to random deadlocks in one of these two tests when
running "make check -j5".
* tests/guix-daemon.sh (tcp_socket): Change port to 9998 to avoid
collision with tests/lint.scm.
Until now the caller couldn't tell the different between a spec like
"foo:out" and one like "foo". This change allows users to distinguish
between these two cases.
* gnu/packages.scm (specification->package+output): Disable output
membership test when OUTPUT = #f and SUB-DRV = #f.
* tests/packages.scm ("specification->package+output")
("specification->package+output invalid output")
("specification->package+output no default output")
("specification->package+output invalid output, no default"): New tests.
* guix/derivations.scm (build-derivations): Accept <derivation-input>
records among DERIVATIONS.
* tests/derivations.scm ("build-derivations with specific output"): Test
it.
Consequently, the whole graph of <derivation> object is readily
available without having to go through 'read-derivation-from-file',
which could have cache misses if the requested <derivation> object had
been GC'd in the meantime. This is an important property for the
performance of things like 'derivation-build-plan' that traverse the
derivation graph.
* guix/derivations.scm (<derivation-input>): Replace 'path' field by
'derivation'.
(derivation-input-path): Adjust accordingly.
(derivation-input-key): New procedure.
(derivation-input-output-paths): Adjust accordingly.
(coalesce-duplicate-inputs): Likewise.
(derivation-prerequisites): Use 'derivation-input-key' to compute keys
for INPUT-SET.
(derivation-build-plan): Likewise.
(read-derivation): Add optional 'read-derivation-from-file' parameter.
[make-input-drvs]: Call it.
(write-derivation)[write-input]: Adjust to new <derivation-input>.
(derivation/masked-inputs): Likewise, and remove redundant
'coalesce-duplicate-inputs' call.
(derivation)[input->derivation-input]: Change to consider only the
derivation case. Update call to 'make-derivation-input'.
[input->source]: New procedure.
Separate sources from inputs.
(map-derivation): Adjust to new <derivation-input>.
* tests/derivations.scm ("parse & export"): Pass a second argument to
'read-derivation'.
("build-expression->derivation and derivation-prerequisites")
("derivation-prerequisites and valid-derivation-input?"): Adjust to new
<derivation-input>.
