Per discussion in https://github.com/martinvonz/jj/discussions/2555. I'm
okay with either way, but it's confusing if we had "branch create" and
"branch set" and both of these could create a new branch.
Renamed `description_template_for_commit` to
`description_template_for_describe` since it's only used in
`cmd_describe`.
Renamed `description_template_for_cmd_split` to
`description_template_for_commit` and modified to accomodate empty
`intro` argument.
Fixes#2439.
As discussed in Discord, it's less useful if remote_branches() included
Git-tracking branches. Users wouldn't consider the backing Git repo as
a remote.
We could allow explicit 'remote_branches(remote=exact:"git")' query by changing
the default remote pattern to something like 'remote=~exact:"git"'. I don't
know which will be better overall, but we don't have support for negative
patterns anyway.
Summary: A natural extension of the existing support, as suggested by Scott
Olson. Closes#2496.
Signed-off-by: Austin Seipp <aseipp@pobox.com>
Change-Id: I91c9c8c377ad67ccde7945ed41af6c79
My gut feeling is that gitoxide aims to be more transparent than libgit2. We'll
need to know more about the underlying Git data model.
Random comments on gix API:
* gix::Repository provides API similar to git2::Repository, but has less
"convenient" functions. For example, we need to use .find_object() +
.try_to/into_<kind>() instead of .find_<kind>().
* gix::Object, Blob, etc. own raw data as bytes. gix::object and gix::objs
types provide high-level views on such data.
* Tree building is pretty low-level compared to git2.
* gix leverages bstr (i.e. bytes) extensively.
It's probably not difficult to migrate git::import/export_refs(). It might
help eliminate the startup overhead of libssl initialization. The gix-based
GitBackend appears to be a bit faster, but that wouldn't practically matter.
#2316
Like "jj log PATHS...", unmatched name isn't an error. I don't think
"jj branch list glob:'push-*'" should fail just because there are no in-flight
PR branches.
Makes sure that, at the point the commit summary for the new commit is written,
the original commit that is being rewritten is already abandoned. Otherwise,
once we show divergent change ids (in a subsequent commit) in the short commit
template, the commits would be shown as divergent.
This also has an effect on whether branches are displayed next to the commit;
the changes in test_resotre_command happen because, now, the branch is properly
propagated to the restored commit before its summary is displayed.
test_templater_alias(), test_templater_alias_override(), and
test_templater_bad_alias_decl() aren't moved since they also test config loading
and error formatting. The first test in test_templater_parse_error() is left for
the same reason. test_templater_upper_lower() depends on the commit templater.
I don't think many of the tests in test_templater.rs should use "jj log" command
as they check very specific template syntax and function behaviors. Let's move
them to in-module tests. We could add a separate test file, but we would have
to export a couple of templater macros.
test_templater_timestamp_method() is migrated as example.
Since "jj git fetch --branch" supports glob patterns, users would expect that
"jj git push --branch glob:.." also works.
The error handling bits are copied from "branch" sub commands. We might want to
extract it to a common helper function, but I haven't figured out a reasonable
boundary point yet.
AFAICT, all callers of `Merge::to_file_merge()` are already well
prepared for working with executable files. It's called from these
places:
* `local_working_copy.rs`: Materialized conflicts are correctly
updated using `Merge::with_new_file_ids()`.
* `merge_tools/`: Same as above.
* `cmd_cat()`: We already ignore the executable bit when we print
non-conflicted files, so it makes sense to also ignore it for
conflicted files.
* `git_diff_part()`: We print all conflicts with mode "100644" (the
mode for regular files). Maybe it's best to use "100755" for
conflicts that are unambiguously executable, or maybe it's better to
use a fake mode like "000000" for all conflicts. Either way, the
current behavior seems fine.
* `diff_content()`: We use the diff content in various diff
formats. We could add more detail about the executable bits in some
of them, but I think the current output is fine. For example,
instead of our current "Created conflict in my-file", we could say
"Created conflict in executable file my-file" or "Created conflict
in ambiguously executable file my-file". That's getting verbose,
though.
So, I think all we need to do is to make `Merge::to_file_merge()` not
require its inputs to be non-executable.
Closes#1279.
I'm about to make conflicts also get materialized in executable
files. We'll lose some of the test coverage in `test_chmod_command.rs`
then, because the those tests rely on the materialized content to
describe the executable bits. So this commit adds a debug command for
printing tree values and uses that in the tests.
If we add glob support, users will probably want to do something like
'jj branch untrack glob:"*@origin"'. It would be annoying if the command
failed just because one of the remote branches has already been untracked.
Since branch tracking/untracking is idempotent, it's safe to continue in
those cases.
The parse rule is lax compared to revset. We could require the pattern to be
quoted, but that would mean glob patterns have to be quoted like 'glob:"foo*"'.
I personally don't mind if "jj branch list" showed all non-tracking branches,
but I agree it would be a mess if ~500 remote branches were listed. So let's
hide them by default as non-tracking branches aren't so interesting.
Closes#1136
This will be the option to include non-tracking remote branches. We could add
more fine-grained filtering flags, but I think --all is good enough and easier
to remember.
This patch also updates many of the test outputs to include synchronized remote
branches. I think verbose outputs will help catch future bugs.
This replaces our existing mechanism of adding `/.jj/` to
`.git/info/exclude` by adding `*` to `.jj/.gitignore`, as suggested by
@ppwwyyxx. That simplifies the code quite a bit, and it avoids the
problem with `.git/info/exclude` not existing (it apparently doesn't
exist when the user uses
https://git-scm.com/docs/git-init#_template_directory).
Closes#2385.
We can provide more actionable error message than "not fast-forwardable". If
the push was fast-forwardable, "jj branch track" should be able to merge the
remote branch without conflicts, so the added step would be minimal.
Although this is logically correct, the error message is a bit cryptic. It's
probably better to reject push if non-tracking remote branches exist.
#1136
This means that the commits previously pinned by remote branches are no longer
abandoned. I think that's more correct since "push" is the operation to
propagate local view to remote, and uninteresting commits should have been
locally abandoned.
Since I'm going to make git::push_branches() update the repo view internally,
it should fail fast if the remote name is reserved. Before, the problem was
detected on git::import_refs().
This patch adds MutableRepo::track_remote_branch() as we'll probably need to
track the default branch on "jj git clone". untrack_remote_branch() is also
added for consistency.
I'm not sure if this is the best way to render non-tracking branches, but
it helps to write CLI tests. Maybe we can add some hint or decoration to
non-tracking branches, but I'd like to avoid bikeshedding at this point.
Since we haven't migrated the push function yet, a deleted branch can be
pushed to non-tracking remotes. This will be addressed later.
#1136
Summary: Yuya's changes and mine had a semantic conflict ("merge skew") between
the two of them, as b7c7b19e changed the `op log `output slightly, whereas
220292ad included a new test that used `op log` itself.
Generated by `cargo insta review`.
Signed-off-by: Austin Seipp <aseipp@pobox.com>
Change-Id: I51d4de7316b1abc09be4f9fa0dd0d1a1
We could fix do_git_clone() instead, but it seemed a bit weird that the
git_repo_path is relative to the store path which is unknown to callers.
Fixes#2374
There's a subtle behavior change. Unlike the original remove_remote_branch(),
remote_views entry is not discarded when the branches map becomes empty. The
reasoning here is that the remote view can be added/removed when the remote
is added/removed respectively, though that's not implemented yet. Since the
serialized data cannot represent an empty remote, such view may generate
non-unique content hash.
Summary: This allows `workspace forget` to forget multiple workspaces in a
single action; it now behaves more consistently with other verbs like `abandon`
which can take multiple revisions at one time.
There's some hoop-jumping involved to ensure the oplog transaction description
looks nice, but as they say: small conveniences cost a lot.
Signed-off-by: Austin Seipp <aseipp@pobox.com>
Change-Id: Id91da269f87b145010c870b7dc043748
Summary: Workspaces are most useful to test different versions (commits) of
the tree within the same repository, but in many cases you want to check out a
specific commit within a workspace.
Make that trivial with a `--revision` option which will be used as the basis
for the new workspace. If no `-r` option is given, then the previous behavior
applies: the workspace is created with a working copy commit created on top of
the current working copy commit's parent.
Signed-off-by: Austin Seipp <aseipp@pobox.com>
Change-Id: I23549efe29bc23fb9f75437b6023c237
Before this patch, it was an error to run `jj config set --user foo
'[1]'` twice. But it's only been broken since the previous commit
because '[1]' was interpreted as a string before then.
Now we have a separate map for "git" tracking remote, we can always preserve
the last imported/exported git_refs. The option to restore git-tracking refs
has been removed. Perhaps, --what can be reorganized as --local and --remote
<NAME>.
Before this patch, when updating to a commit that has a file that's
currently an ignored file on disk, jj would crash. After this patch,
we instead leave the conflicting files or directories on disk. We
print a helpful message about how to inspect the differences between
the intended working copy and the actual working copy, and how to
discard the unintended changes.
Closes#976.
On my Debian laptop, openssl_init() takes ~30ms to load the default CA
certificates serialized in PEM format, and the cost is added to each jj
invocation. This change saves 20s (of 50s) on my machine.
% wc -l /usr/lib/ssl/cert.pem
3517 /usr/lib/ssl/cert.pem
I've added a boolean flag to the store to ensure that the migration never runs
more than once after the view gets "op restore"-d. I'll probably reorganize the
branches structure to support non-tracking branches later, but updating the
storage format in a single commit would be too involved.
If jj is downgraded, these "git" remote refs would be exported to the Git repo.
Users might have to remove them manually.
I have used the tree-level conflict format for several weeks without
problem (after the fix in 51b5d168ae). Now - right after the 0.10.0
release - seems like a good time to enable the config by default.
I enabled the config in our default configs in the CLI crate to reduce
impact on tests (compared to changing the default in `settings.rs`).
As we can set HEAD to an arbitrary ref by using .reference_symbolic(), we don't
have to manage a ref that can also be valid as a branch name.
Fixes#1495
Apparently, it gets too verbose if the remote history is actively rewritten.
Let's summarize the output for now. The plan is to show the list of moved refs
instead of the full list of abandoned commits.
The problem is that the first non-working-copy commit moves the unborn current
branch to that commit, but jj doesn't "export" the moved branch. Therefore,
the next jj invocation notices the "external" ref change, which was actually
made by jj.
I'm not sure why we play nice by setting the "current" HEAD, but I *think* it's
okay to set the "new" HEAD and reset to the same commit to clear Git index.
This will probably help to understand why you've got conflicts after fetching.
Maybe we can also report changed local refs.
I think the stats should be redirected to stderr, but we have many other similar
messages printed to stdout. I'll probably fix them all at once later.
I think most users who change the set of immutable heads away from
`trunk() | tags()` are going to also want to change the default log
revset to include the newly mutable commit and to exclude the newly
immutable commits. So let's update the default log revset to use
`immutable_heads()` instead.
`test_templater` changed because we have overridden the set of
immutable commits there so `jj log` now includes the remote branch.
This adds a new `revset-aliases.immutable_heads()s` config for
defining the set of immutable commits. The set is defined as the
configured revset, as well as its ancestors, and the root commit
commit (even if the configured set is empty).
This patch also adds enforcement of the config where we already had
checks preventing rewrite of the root commit. The working-copy commit
is implicitly assumed to be writable in most cases. Specifically, we
won't prevent amending the working copy even if the user includes it
in the config but we do prevent `jj edit @` in that case. That seems
good enough to me. Maybe we should emit a warning when the working
copy is in the set of immutable commits.
Maybe we should add support for something more like [Mercurial's
phases](https://wiki.mercurial-scm.org/Phases), which is propagated on
push and pull. There's already some affordance for that in the view
object's `public_heads` field. However, this is simpler, especially
since we can't propagate the phase to Git remotes, and seems like a
good start. Also, it lets you say that commits authored by other users
are immutable, for example.
For now, the functionality is in the CLI library. I'm not sure if we
want to move it into the library crate. I'm leaning towards letting
library users do whatever they want without being restricted by
immutable commits. I do think we should move the functionality into a
future `ui-lib` or `ui-util` crate. That crate would have most of the
functionality in the current `cli_util` module (but in a
non-CLI-specific form).
I'm going to make this function check against a configurable revset
indicating immutable commits. It's more efficient to do that by
evaluating the revset only once.
We may want to have a version of the function where we pass in an
unevaluated revset expression. That would allow us to error out if the
user accidentally tries to rebase a large set of commits, without
having to evaluate the whole set first.
Once we add support for immutable commits, `jj duplicate` should be
allowed to create duplicate of them. The reason it can't duplicate the
root commit is that it would mean there would be multiple root
commits, which would break the invariant that the single root commit
is the only root commit (and the backends refuse to write a commit
without parents). So let's have `jj duplicate` check specifically that
the user doesn't try to duplicate the root commit instead.
Since we have overloaded operator symbols, we need to deduplicate them
upfront. Legacy and compat operators are also removed from the suggestion.
It's a bit ugly to mutate the error struct before calling Error::renamed_rule(),
but I think it's still better than reimplementing message formatting function.
As we discussed in #1928, it seems better to print information about
abandoned commits in the context of the pre-abandon state. For
example, that means that we'll include any branches that pointed to
the now-abandoned commits.
This is a naive implementation, which cannot deal with multiple children
or parents stemming from merges.
Note: I gave each command separate a separate argument struct
for extensibility.
Fixes#878
Suppose "x::y" is the operator that defaults to "root()::visible_heads()"
respectively, "::" is identical to "all()". Since we've just changed the
behavior of "..y", ".." is now "root()..visible_heads()" meaning "~root()".
In `LockedWorkingCopy::drop()`, we panic if the caller had not called
`finish()`. IIRC, the idea was both to find bugs where we forgot to
call `finish()` and to prevent continuing with a modified
`WorkingCopy` instance. I don't think the former has been a problem in
practice. It has been a problem in practice to call `discard()` to
avoid the panic, though. To address that, we can make the `Drop`
implementation discard the changes (forcing a reload of the state if
the working copy is accessed again).
I also converted the error from `InternalError` to `UserError`. So far
I've intented to use `InternalError` only to indicate bugs or corrupt
repos. I'm not sure that's a good idea, and we can revisit it later.
If the path is too long to fit on the screen, this patch makes it so
we elide the first part of it. It goes a bit further and trims it down
to ~70% of the screen, giving some room for the stat. This seems
somewhat similar to what Git does.
`insta` ignores leading indentation (as long as it's consistent within
the snapshot), but when a test case fails and you let `cargo insta`
update it, it's going to use a specific indentation. There were a few
tests that didn't match that indentation, which could lead to
surprising diffs if the tests fail at some point.
This shows that there's too much padding because we pad based on
number of bytes.
I had to reduce the path names for the file names not to get too long
for my file system.
We can still crash on terminals that are less than 4 characters wide
(maybe it doesn't matter if we do because the user can't tell the
crash report from a diffstat in such a terminal?). This patch fixes
the crash.
We would run into a panic due to "attempt to subtract with overflow"
if the path was long. This patch fixes that and adds tests showing the
current behavior when there are long paths and/or large diffs.
When we start writing tree-level conflicts in an existing repo, we
don't want commits that change the format to be non-empty if they
don't change any content. This patch updates `MergeTreeId::eq()` to
consider two resolved trees equal even if only their `MergedTreeId`
variant is different (one is path-level and one is tree-level).
I think I've gone through all places we compare tree ids and checked
that it's safe to compare them this way. One consequence is that
rebasing a commit without changing the parents (typically
auto-rebasing after `jj describe`) will not lead to the tree id
getting upgraded, due to an optimization we have for that case. I
don't think that's serious enough to handle specially; we'll have to
support the old format for existing repos for a while regardless of a
few commits not getting upgraded right away.
The number of failing tests with the config option enabled drop from
108 to 11 with this patch.
We're finally ready to start writing trees using the new format where
we represent conflicts by having multiple trees in the commit instead
of having a single tree with multiple entries at a path. This patch
adds a config option for that. It's not ready to be used yet, so I
haven't updated the release notes or other documentation.
I added only a simple CLI test for testing what happens when the
config is enabled in an existing repo. 108 tests currently fail if we
flip the default.
With the idea that less severe placeholders (like description) could
(and should) explicitly "opt out".
(Both email and name placeholders will be red with this change.)
I made it simply fail on explicit fetch/import, and ignored on implicit import.
Since the error mode is predictable and less likely to occur. I don't think it
makes sense to implement warning propagation just for this.
Closes#1690.
This switches the whole `diff_util` module to working with
`MergedTree`, `Merge<Option<TreeValue>>` etc., so it can support
tree-level conflicts.
Since we want to avoid using `ConflictId`s, I switched the hash we use
for conflicts in `--git` style diffs to use an all-'0' id instead of
using the conflict id.
This patch also extracts format_detailed_signature() function to deduplicate
the "show" template bits.
The added placeholder templates aren't labeled as "empty". If needed, I think
the whole template can be labeled as "empty" (or "empty_commit") just like
"working_copy".
Closes#2112
I think this is less surprising than falling back to the default length.
i64-to-usize conversion can also overflow on 32 bit environment, but I'm not
bothered to handle overflow scenario.
This allows negative numbers, which also means functions which took numbers can now take negative numbers
Luckily, they all already handled this exactly as expected.
The way `jj git push` without arguments chooses branches pointing to
either `@` or `@-` is unusual and difficult to explain. Now that we
have `-r`, we could instead default it to `-r '@-::@'`. However, I
think it seems likely that users will want to push all local branches
leading up to `@` from the closest remote branch. That's typically
what I want. This patch changes the default to do that.
If there are branches in the revset that don't need to be pushed
because they already match the destination, we currently just print
`Nothing changed.` It seems consistent with that to also treat it as
success if there are no branches in the specified set to start
with. This patch makes the command print a warning in that case
instead.
We resolve file paths into repo-relative paths while parsing the
revset expression, so I think it's consistent to also resolve which
workspace "@" refers to while parsing it. That means we won't need the
workspace context both while parsing and while resolving symbols.
In order to break things like `author("martinvonz@")` (thanks to @yuja
for catching this), I also changed the parsing of working-copy
expressions so they are not allowed to be
quoted. `author(martinvonz@)` will therefore be an error now. That
seems like a small improvement anyway, since we have recently talked
about making `root` and `[workspace]@` not parsed as other symbols.
This commit replaces the functions `UserSettings::user_name_placeholder()`` and
`UserSettings::user_email_placeholder()` with `const` `&str`s to emphasize that
the placeholder strings must not be changed to support commits without
names or email addresses made before this change.
Bright green really pops on my screen, and I don't think there is a reason
for the root commit to be attention-grabbing.
This follows up on https://github.com/martinvonz/jj/pull/2084.
The syntax is slightly different from Mercurial. In Mercurial, a pattern must
be quoted like "<kind>:<needle>". In JJ, <kind> is a separate parsing node, and
it must not appear in a quoted string. This allows us to report unknown prefix
as an error.
There's another subtle behavior difference. In Mercurial, branch(unknown) is
an error, whereas our branches(literal:unknown) is resolved to an empty set.
I think erroring out doesn't make sense for JJ since branches() by default
performs substring matching, so its behavior is more like a filter.
The parser abuses DAG range syntax for now. It can be rewritten once we remove
the deprecated x:y range syntax.
One use case for `jj split` is when creating a new commit from some of
the changes in the working copy. If there's no description on the
working-copy commit in that case, it seems better to not ask the user
to provide one when they're splitting the commit either.
I've extracted the `builtin_log_root` template for users to customize the
default templates without fully overriding them, for example I would remove
the change_id/commit_id for myself - and we discussed in Discord that leaving
those makes sense for the user to be reminded/teached that the root commit has
a change id made from z's.
This basically means that heads in a filtered graph appear in reverse
chronological order. Before, "jj log -r 'tags()'" in linux-stable repo would
look randomly sorted once you ran "jj debug reindex" in it.
With this change, indexing is more like breadth-first search, and BFS is
known to be bad at rendering nice graph (because branches run in parallel.)
However, we have a post process to group topological branches, so we don't
have this problem. For serialization formats like Mercurial's revlog iirc,
BFS leads to bad compression ratio, but our index isn't that kind of data.
Reindexing gets slightly slower, but I think this is negligible.
(in Git repository)
% hyperfine --warmup 3 --runs 10 "jj debug reindex --ignore-working-copy"
(original)
Time (mean ± σ): 1.521 s ± 0.027 s [User: 1.307 s, System: 0.211 s]
Range (min … max): 1.486 s … 1.573 s 10 runs
(new)
Time (mean ± σ): 1.568 s ± 0.027 s [User: 1.368 s, System: 0.197 s]
Range (min … max): 1.531 s … 1.625 s 10 runs
Another idea is to sort heads chronologically and run DFS-based topological
sorting. It's ad-hoc, but worked surprisingly well for my local repositories.
For repositories with lots of long-running branches, this commit will provide
more predictable result than DFS-based one.
`jj chmod` won't operate on conflicts involving non-files on the
positive sides. However, the error message says "None of the sides of
the conflict are files", which is not correct.
That is, jj will use ui.default_description as a starting point when
user is about to describe an empty change.
I think it might be confusing to do this with -m / --stdin (violates
WYSIWYG), so I'm only doing this when jj invokes an editor.
Also, this could evolve into a proper template in the future instead of
just plain text, to allow inheriting from parent change(s), for example.
Partially addresses #1354.
We anyway trim the newlines eventually and this just does that eagerly
so we output the "correct" description back to stdout (on describe for
example, we'd now print the first non empty line).
Empty files can be confusing in diff output. For example:
```
Added regular file file1:
Added regular file file2:
1: foo
```
This commit adds an "(empty)" placeholder instead. Since it's not
colored, and doesn't have line numbers, it will hopefully not be
mistaken for a file with the contents "(empty)".
AFAIK, we can't make HEAD detached in an empty Git repository, so we need
to temporarily switch to the new default branch before checking out.
Fixes#2047
They are shown next to the change and commit id, since they are other names the
commit can be referred by.
The description is separated from the branches by a ` | ` when there are
branches, so that one can tell the branches from the description without color.
The result looks like this: 
Maybe we could load GitBackend without resolving .git symlink, but that would
introduce more subtle bugs. Instead, we calculate the expected Git workdir path
from the canonical ".git" path.
Fixes#2011
