aboutsummaryrefslogtreecommitdiff
path: root/src/libfetchers/git.cc
AgeCommit message (Collapse)Author
2022-12-07Trivial changes from the lazy-trees branchEelco Dolstra
2022-10-14Fix #7146Andrew Brooks
When fetching a non-local git repo by ref (and no rev), don't consider unrelated cached revs for the same repository.
2022-08-17Don't pass --force to 'git add'Eelco Dolstra
Fixes #5810.
2022-06-26fetchGit: document `shallow` argumentGuillaume Girol
2022-06-10libfetchers/git: add missing `--git-dir` flagsNaïm Favier
2022-05-27Fix a segfault in the git fetcherThéophane Hufschmitt
The git fetcher code used to dereference the (potentially empty) `ref` input attribute. This was magically working, probably because the compiler somehow outsmarted us, but is now blowing up with newer nixpkgs versions. Fix that by not trying to access this field while we don't know for sure that it has been defined. Fix #6554
2022-05-24libfetchers: drop `getGitDir` and hardcode `.git`Maximilian Bosch
As discussed[1] this is most likely not desirable. [1] https://github.com/NixOS/nix/pull/6440#issuecomment-1120876248
2022-05-04Fix the parsing of the sourcehut refs fileThéophane Hufschmitt
Since a26be9f3b89be2ee90c6358250b9889b37f95cf8, the same parser is used to parse the result of sourcehut’s `HEAD` endpoint (coming from [git dumb protocol]) and the output of `git ls-remote`. However, they are very slightly different (the former doesn’t specify the current reference since it’s implied to be `HEAD`). Unify both, and make the parser a bit more robust and understandable (by making it more typed and adding tests for it) [git dumb protocol]: https://git-scm.com/book/en/v2/Git-Internals-Transfer-Protocols#_the_dumb_protocol
2022-05-02Style fixesEelco Dolstra
2022-05-02Merge pull request #6470 from Ma27/git-followupEelco Dolstra
libfetchers/git: fix every occasion of a permission error
2022-04-30libfetchers/git: fix every occasion of a permission errorMaximilian Bosch
I'm afraid I missed a few problematic `git(1)`-calls while implementing PR #6440, sorry for that! Upon investigating what went wrong, I realized that I only tested against the "cached"-case by accident because my git-checkout with my system's flake was apparently cached during my debugging. I managed to trigger the original issue again by running: $ git commit --allow-empty -m "tmp" $ sudo nixos-rebuild switch --flake .# -L --builders '' Since `repoDir` points to the checkout that's potentially owned by another user, I decided to add `--git-dir` to each call affecting `repoDir`. Since the `tmpDir` for the temporary submodule-checkout is created by Nix itself, it doesn't seem to be an issue. Sorry for that, it should be fine now.
2022-04-29Extract git reference parsing to a shared libraryKjetil Orbekk
These utility functions can be shared between the git and github fetchers.
2022-04-29Store cached head in cached git repoKjetil Orbekk
The previous head caching implementation stored two paths in the local cache; one for the cached git repo and another textfile containing the resolved HEAD ref. This commit instead stores the resolved HEAD by setting the HEAD ref in the local cache appropriately.
2022-04-29Refactor fetching of dirty workdirKjetil Orbekk
Extract the handling of a local dirty workdir to a helper function.
2022-04-29Resolve reference for remote repositoryKjetil Orbekk
Resolves the HEAD reference from the remote repository instead of assuming "master".
2022-04-24libfetchers/git: fix for nixos-rebuildMaximilian Bosch
The `--git-dir=` must be `.` in some cases (for cached repos that are "bare" repos in `~/.cache/nix/gitv3`). With this fix we can add `--git-dir` to each `git`-invokation needed for `nixos-rebuild`.
2022-04-23libfetchers/git: hardcode `--git-dir`Maximilian Bosch
To demonstrate the problem: * You need a `git` at 2.33.3 in your $PATH * An expression like this in a git repository: ``` nix { outputs = { self, nixpkgs }: { packages.foo.x86_64-linux = with nixpkgs.legacyPackages.x86_64-linux; runCommand "snens" { } '' echo ${(builtins.fetchGit ./.).lastModifiedDate} > $out ''; }; } ``` Now, when instantiating the package via `builtins.getFlake`, it fails on Nix 2.7 like this: $ nix-instantiate -E '(builtins.getFlake "'"$(pwd)"'").packages.foo.x86_64-linux' fatal: unsafe repository ('/nix/store/a7j3125km4h8l0p71q6ssfkxamfh5d61-source' is owned by someone else) To add an exception for this directory, call: git config --global --add safe.directory /nix/store/a7j3125km4h8l0p71q6ssfkxamfh5d61-source error: program 'git' failed with exit code 128 (use '--show-trace' to show detailed location information) This breaks e.g. `nixops`-deployments using flakes with similar expressions as shown above. The cause for this is that `git(1)` tries to find the highest `.git`-directory in the directory tree and if it finds a such a directory, but with another owning user (root vs. the user who evaluates the expression), it fails as above. This was changed recently to fix CVE-2022-24765[1]. By explicitly specifying `--git-dir`, Git assumes to be in the top-level directory and doesn't attempt to look for a `.git`-directory in the parent directories and thus the code-path leading to said error is never reached. [1] https://lore.kernel.org/git/xmqqv8veb5i6.fsf@gitster.g/
2022-04-14Merge pull request #6387 from Uthar/fixEelco Dolstra
assert hash types for Git and Mercurial
2022-04-12assert hash types for Git and MercurialKasper Gałkowski
2022-04-09libfetchers: Replace regex to clarify intentDaniel Pauls
2022-04-06libfetchers: Fix assertionDaniel Pauls
The filter expects all paths to have a prefix of the raw `actualUrl`, but `Store::addToStore(...)` provides absolute canonicalized paths. To fix this create an absolute and canonicalized path from the `actualUrl` and use it instead. Fixes #6195.
2022-03-11StyleEelco Dolstra
2022-03-11Merge pull request #5758 from mschwaig/fix-git-workspace-dirty-detectionThéophane Hufschmitt
git fetcher: fix detection of dirty git workspaces
2022-03-01Move some stuff from `Settings` to a new `FetchSettings`.John Ericson
Starting work on #5638 The exact boundary between `FetchSettings` and `EvalSettings` is not clear to me, but that's fine. First lets clean out `libstore`, and then worry about what, if anything, should be the separation between those two.
2022-02-24libfetchers: Rename immutable -> lockedEelco Dolstra
This is more consistent with flake terminology.
2022-02-19git fetcher: set locale for rev-parseMartin Schwaighofer
2022-02-19git fetcher: distinguish errors more preciselyMartin Schwaighofer
2022-02-19git fetcher: invoke diff instead of diff-indexMartin Schwaighofer
diff-index operates on the view that git has of the working tree, which might be outdated. The higher-level diff command does this automatically. This change also adds handling for submodules. fixes #4140 Alternative fixes would be invoking update-index before diff-index or matching more closely what require_clean_work_tree from git-sh-setup.sh does, but both those options make it more difficult to reason about correctness.
2022-02-19git fetcher: improve check for valid repositoryMartin Schwaighofer
The .git/refs/heads directory might be empty for a valid usable git repository. This often happens in CI environments, which might only fetch commits, not branches. Therefore instead we let git itself check if HEAD points to something that looks like a commit. fixes #5302
2022-02-16InputScheme::fetch(): Return a StorePath instead of a TreeEelco Dolstra
2021-11-16Merge pull request #5452 from Kha/git-extend-lockEelco Dolstra
git: extend cache dir lock over all mutating operations
2021-11-01git: extend cache dir lock over all mutating operationsSebastian Ullrich
2021-10-26Support building flakes from a Git repo url with submodules query parameterTimothy
2021-09-24Explicitly set initial branch name for gitoxalica
2021-09-22Revert "Merge pull request #4922 from nrdxp/default-submodules"Eelco Dolstra
This reverts commit 6678e98411cd3bfe8143a4ba1b35d1451babce32, reversing changes made to 90b2dd570cbd8313a8cf45b3cf66ddef2bb06e07.
2021-09-14Fix macOS buildEelco Dolstra
2021-09-13RunOptions: Use designated initializersEelco Dolstra
Also get rid of _killStderr because it wasn't actually checked anywhere.
2021-09-07Merge pull request #4922 from nrdxp/default-submodulesShea Levy
libfetchers/git: fetch submodules by default
2021-08-02fix git init race conditionMichael Fellinger
2021-07-06Allow `fetchGit` to take a `name` argumentregnat
Fix #3388
2021-06-17libfetchers/git: fetch submodules by defaultTimothy DeHerrera
2021-04-19Include sys/wait.h everywhere WIFEXITED etc is usedAlyssa Ross
This is required on NetBSD, and I think FreeBSD too.
2021-03-30fetchGit: don't prefix "refs/heads/" on ref = "HEAD"Bjørn Forsman
This fixes builtins.fetchGit { url = ...; ref = "HEAD"; }, that works in stable nix (v2.3.10), but is broken in nix master: $ ./result/bin/nix repl Welcome to Nix version 2.4pre19700101_dd77f71. Type :? for help. nix-repl> builtins.fetchGit { url = "https://github.com/NixOS/nix"; ref = "HEAD"; } fetching Git repository 'https://github.com/NixOS/nix'fatal: couldn't find remote ref refs/heads/HEAD error: program 'git' failed with exit code 128 The documentation for builtins.fetchGit says ref = "HEAD" is the default, so it should also be supported to explicitly pass it. I came across this issue because poetry2nix can use ref = "HEAD" in some situations. Fixes #4674.
2021-03-01Add support for bare git repositories with git+fileKjetil Orbekk
Local git repositories are normally used directly instead of cloning. This commit checks if a repo is bare and forces a clone. Co-authored-by: Théophane Hufschmitt <regnat@users.noreply.github.com>
2020-12-22Add explicit `allRefs = true;` argument to `fetchGit`Maximilian Bosch
Sometimes it's necessary to fetch a git repository at a revision and it's unknown which ref contains the revision in question. An example would be a Cargo.lock which only provides the URL and the revision when using a git repository as build input. However it's considered a bad practice to perform a full checkout of a repository since this may take a lot of time and can eat up a lot of disk space. This patch makes a full checkout explicit by adding an `allRefs` argument to `builtins.fetchGit` which fetches all refs if explicitly set to true. Closes #2409
2020-12-22Provide a more meaningful error-message for `builtins.fetchGit` if a ↵Maximilian Bosch
revision can't be checked out A common pitfall when using e.g. `builtins.fetchGit` is the `fatal: not a tree object`-error when trying to fetch a revision of a git-repository that isn't on the `master` branch and no `ref` is specified. In order to make clear what's the problem, I added a simple check whether the revision in question exists and if it doesn't a more meaningful error-message is displayed: ``` nix-repl> builtins.fetchGit { url = "https://github.com/owner/myrepo"; rev = "<commit not on master>"; } moderror: --- Error -------------------------------------------------------------------- nix Cannot find Git revision 'bf1cc5c648e6aed7360448a3745bb2fe4fbbf0e9' in ref 'master' of repository 'https://gitlab.com/Ma27/nvim.nix'! Please make sure that the rev exists on the ref you've specified or add allRefs = true; to fetchGit. ``` Closes #2431
2020-11-10Fix use of dirty Git/Mercurial inputs with chroot storesEelco Dolstra
Fixes: $ nix build --store /tmp/nix /home/eelco/Dev/patchelf#hydraJobs.build.x86_64-linux warning: Git tree '/home/eelco/Dev/patchelf' is dirty error: --- RestrictedPathError ------------------------------------------------------------------------------------------- nix access to path '/tmp/nix/nix/store/xmkvfmffk7xfnazykb5kx999aika8an4-source/flake.nix' is forbidden in restricted mode (use '--show-trace' to show detailed location information)
2020-10-06Remove static variable name clashesEelco Dolstra
This was useful for an experiment with building Nix as a single compilation unit. It's not very useful otherwise but also doesn't hurt...
2020-09-21Don't include <regex> in header filesEelco Dolstra
This reduces compilation time by ~15 seconds (CPU time). Issue #4045.
2020-08-13Don't try to parse signature check as commit timestampChuck
When the log.showSignature git setting is enabled, the output of "git log" contains signature verification information in addition to the timestamp GitInputScheme::fetch wants: $ git log -1 --format=%ct gpg: Signature made Sat 07 Sep 2019 02:02:03 PM PDT gpg: using RSA key 0123456789ABCDEF0123456789ABCDEF01234567 gpg: issuer "user@example.com" gpg: Good signature from "User <user@example.com>" [ultimate] 1567890123 1567890123 For folks that had log.showSignature set, this caused all nix operations on flakes to fail: $ nix build error: stoull