This allows us to select rendering function hunk by hunk. For example, a hunk
with lots of small changes could be rendered without interleaving left/right
words. Another good thing is that context line handling can be simplified as
the whole context hunk is available.
I'm going to split color-words diffs to by_line() and by_word() stages.
Perhaps, Diff::default_refinement() can be removed once all non-test callers
are migrated.
The code in both cli/src/commands/{next,prev}.rs is identical except
for the direction of movement. This commit pull the parts that make
sense out into cli/src/movement_util.rs so it's easier to see the
differences.
Part of #3947
Add gratuitous `jj log` output to more points in the tests.
This makes it easier to understand the intended changes
by literally visualizing the commit tree we are after each movement.
This is at least useful for me since I find the new+squash workflow
confusing.
Test behaviour is not changed.
Part of #3947
I'm thinking of adding some heuristics to render hunks containing lots of
small word changes differently, in a similar manner to the unified diffs. This
patch might help add some pre/post-processing at consumer.
files::diff() is inlined to caller to get around 'self borrowing.
When rebasing a new child commit on top of the moved commit(s), the
order of the new child commit's parent commits is now correctly
preserved if the original parent commit is now a parent of the moved
commit(s).
Closes#3969.
I think they were adding too much noise to commit diffs. Only the tests
focused on skipping rebasing will include the commit and change IDs,
other tests will omit them.
* Add `builtin_immutable_heads()` in the `revsets.toml`.
* Redefine `immutable_heads()` in terms of `builtin_immutable_heads()`
* Warn if user redefines `builtin_immutable_heads()`, `mutable()` or
`immutable()`.
* Update module constant in revset_util.rs from BUILTIN_IMMUTABLE_HEADS
to USER_IMMUTABLE_HEADS to avoid confusion since it points at
`immutable_heads()` **and** we now have a revset-alias
literally named `builtin_immutable_heads()`.
* Add unittest
* Update CHANGELOG
* Update documentation.
Fixes: #4162
- add support for copy tracking to `diff --stat`
- switch `--summary` to match git's output more closely
- rework `show_diff_summary` signature to be more consistent
Add home directory expansion for SSH key filepaths. This allows the
`signing.key` configuration value to work more universally across both
Linux and macOS without requiring an absolute path.
This moved and renamed the previous `expand_git_path` function to a more
generic location, and the prior use was updated accordingly.
- force each diff command to explicitly enable copy tracking
- enable copy tracking in diff_summary
- post-process for diff iterator
- post-process for diff stream
- update changelog
- use a single commit instead of an array of them. This simplifies the
implementation. A higher level api can wrap this when an array of
commits is desired and those semantics are figured out.
- since this API is directly 1-1 on parents, there are no conflicts
- if we introduce a higher level API that handles lists of commits, we
may need to restore the conflict/resolved distinction, but for now
simplify
The "git" command appears to chdir() to the --work-tree directory first, then
read() the core.excludesFile file. There's no manual relative path resolution
in "git".
Fixes#4222
Since "op abandon" just rewrites DAG, it works no matter if the heads are
merged or not. This change will help crash recovery. "op abandon
--at-op=<one-of-the-heads>" can't be used because ancestor operations would be
preserved by the other head.
Suppose a squash node in obslog is analogous to a merge in revisions log, it
makes sense to show diffs from auto-merge (or auto-squash) parents. This
basically means a non-partial squash node no longer shows diffs.
This also fixes missing diffs at the root predecessors if there were.
As the doc comment states, this function only supports change of "a single
added and removed commit", "only a single added", or "single removed commit."
It doesn't make sense to show diffs from the parent commit ignoring multiple
removed commits for example.
I think it might be nice to have this in the upcoming release, but I'd
like to warn people that their changes will be lost if they aren't
careful, and to not rely on the syntax being fixed just yet.
This partially reverts 543036c753 "cli: run 'op log' without loading repo or
merging concurrent ops." User can now get around the issue by --at-op=@
--ignore-working-copy.
The idea is that --at-op specifies a certain operation, so --at-op=@ can be
interpreted as the option to select _the_ known head operation. This helps
eliminate special cases from "op log" which doesn't snapshot nor merge
concurrent ops.
Author dates and committer dates can be filtered like so:
committer_date(before:"1 hour ago") # more than 1 hour ago
committer_date(after:"1 hour ago") # 1 hour ago or less
A date range can be created by combining revsets. For example, to see any
revisions committed yesterday:
committer_date(after:"yesterday") & committer_date(before:"today")
To avoid always printing the rebase instructions to fix a conflict
even when a child commit to fix the conflict already exists, implement
the following:
* If working commit has conflicts:
* Continue printing the same message we print today.
* If working commit has no conflicts:
* If any parent has conflicts, we print: "Conflict in parent is resolved in working copy".
Also explicitly not printing the "conflicting parent" here, since a merge commit
could have conflict in multiple parents.
* If no parent has any conflicts: exit quietly.
* Update unittests for conflict hinting update.
* Update CHANGELOG
It's not so important, but this removes duplicated "diff" labels from template
output. Perhaps, this also fixes "diff access-denied" label in file-by-file
external diffs.
The inner show_*() functions no longer add "diff" labels, but that's okay
because all CLI callers (except for the templater) use DiffRenderer.
The added test shows the "diff" label is repeated because of auto-labeling of
templater. The original "--color=always" test is also kept to ensure that color
sequences are unchanged even if we remove one of the "diff" labels.
The width parameter is mandatory so it wouldn't fall back to ui.term_width() by
mistake. The API is getting messy and we might want to extract some parameters
to separate struct.
Fixes#4158
Suppose cmd_root() will gain options to print other kind of root paths, it's
probably simpler to inline path handling there instead of turning cmd_root()
into a dispatcher function. If cmd_root() remains with the current form, maybe
we can add aliases.root = ["workspace", "root"] instead.
Users may try to run `jj workspace add <name>` without specifying a
path, which results in the workspace being created in the current
directory. This can be confusing, since the workspace contents will also
be snapshotted in the original workspace if it is not sparse. Adding a
warning should reduce confusion in this case.
The high level changes include:
- Reworking `fix_file_ids()` to loop over multiple candidate tools per file,
piping file content between them. Only the final file content is written to
the store, and content is no longer read for changed files that don't match
any of the configured patterns.
- New struct `ToolsConfig` to represent the parsed/validated configuration.
- New function `get_tools_config()` to create a `ToolsConfig` from a `Config`.
- New tests; the only old behavior that has changed is that we don't require
`fix.tool-command` if `fix.tools` defines one or more tools. The general
approach to validating the config is to fail early if anything is weird.
Co-Authored-By: Josh Steadmon <steadmon@google.com>
This allows tests to easily distinguish the effects of multiple formatters that
could be applied during a single `jj fix` command, which is helpful for testing
the feature for configuring multiple formatters on potentially overlapping
filesets.
Uppercase is a counterpart to lowercase. Append exposes the number of changes
and their order of execution in the file content, which provides a terse way of
writing test expectations.
This implements a building block of "signed-off-by line" #1399 and "commit
--verbose" #1946. We'll probably need an easy way to customize the diff part,
but I'm not sure if it can be as simple as a template alias function. User
might want to embed diffs without "JJ: " prefixes?
Perhaps, we can deprecate "ui.default-description", but it's not addressed in
this patch. It could be replaced with "default_description" template alias,
but we might want to configure default per command. Suppose we add a default
"backout_description" template, it would have to be rendered against the
source commit, not the newly-created backout commit.
The template key is named as "draft_commit_description" because it is the
template to generate an editor template. "templates.commit_description_template"
sounds a bit odd.
There's one minor behavior change: the default description is now terminated
by "\n".
Closes#1354
description_template_for_*() functions will be merged soon.
We don't need to set the default description to the second commit of the split
because its description should be kept empty if it was.
This could be handled by description_template_for_describe(), but I think it's
better to do mutation by caller. I also think commands like "backout" or "new"
can do a similar thing to generate a default commit description from the source
or merge parent commits.
This will help consolidate description_template_for_*() functions and port them
to templater. This change also means the committer (and operation start)
timestamp is set earlier. I think that's good.
This is part of migrating away from legacy trees (with path-level
conflicts). I can't think of any practical impact (we already compare
the tree ids equal).
This basically reverts 20eb9ecec1 "git: don't abandon HEAD commit when it
loses a branch." I think the new behavior is more consistent because the Git
HEAD is equivalent to @- in jj, so it shouldn't be considered a named ref.
Note that we've made old HEAD branch not considered at 92cfffd843 "git: on
external HEAD move, do not abandon old branch."
#4108
It's wrong to deduce loaded_at_head from command-line arguments if the repo was
loaded at an arbitrary operation. Instead, the caller should specify whether
the working copy state is synchronized with the repo view.
I think .for_loaded_repo(ui, workspace, repo, true) would be bad for
readability, so I added separate functions. I'm not happy with the name
.for_temporary_repo(), but it seems okay for the current call sites.
--at-op should be invalid on repo initialization. "init --ignore-working-copy"
could be supported by using working_copy.reset(), but I don't think it's
worth the effort. If the working directory is empty, --ignore-working-copy
is meaningless, and if the directory is not empty, the user would probably
want to do snapshot at some point.
I'm going to fix misuse of CommandHelper::for_loaded_repo(), which expects
that the given repo respects the --at-operation option.
I don't think all of the added tests are useful, but "clone
--ignore-working-copy" might be legit as a replacement for bare repos.
The user can define the setting `git.private-commits` as they desire. For
example:
git.private-commits = 'description(glob:"wip:*")'
If any commits are in this revset, then the push is aborted.
If a commit would be private but already exists on the remote, then it does
not block pushes, nor do its descendents block pushes unless they are also
contained in `git.private-commits`.
Closes#3376
The user probably would expect the path to be relative to their current
directory rather than the workspace root. For instance, if the user is
in a child directory and runs `jj workspace add ../../name`, then they
might be surprised if we printed "../name" instead of "../../name".
These functions mirror the ones in WorkspaceCommandHelper. I'm not sure if all
of them will have to be public, but parse_commit_template() might be useful in
order to implement description template.
Prevents a warning from being printed when renaming branches in a
colocated repo, since git tracking branches were being considered as
remote tracking branches.
It's inconvenient that we have to quote glob patterns as 'glob:"*.rs"'. Suppose
filesets are usually specified in shell, it's better to allow unquoted strings
if possible. This change also means we'll probably abandon #2101 "make the
parsing of string arguments stricter."
Note that we can no longer introduce ? operator or [] subscript syntax in
filesets.
Closes#4053
Now that `jj move` does not accept `-r` (since it has `--from` and `--to`), `jj set --to` seems more useful
than `jj set -r`. `create --to` is also added for ease of switching
between `branch create` and `branch move`.
This patch adds TreeDiff template type to host formatting options. The main
reason of this API design is that diff formats have various incompatible
parameters, so a single .diff(files, format[, options..]) method would become
messy pretty quickly. Another reason is that we can probably add custom
summary templating support as diff.files().map(|file| file.path()..).
RepoPathUiConverter is passed to templater explicitly because the one stored
in RevsetParseContext is behind Option<_>.
This will help remove lifetimed &dyn Repo from diff object in templater.
Function arguments are reordered in a way that all show_*() functions have
common parameters in the same order.
Since fileset and revset languages are syntactically close, we can reparse
revset expression as a fileset. This might sound a bit scary, but helps
eliminate nested quoting like file("~glob:'*.rs'"). One oddity exists in alias
substitution, though. Another possible problem is that we'll need to add fake
operator parsing rules if we introduce incompatibility in fileset, or want to
embed revset expressions in a fileset.
Since "file(x, y)" is equivalent to "file(x|y)", the former will be deprecated.
I'll probably add a mechanism to collect warnings during parsing.
Currently, when there is a commit with two predecessors, the graph
splits into two branches, and all of the predecessors on the first
branch are printed before all of the predecessors on the second branch.
This causes the graph to grow wider with each squashed commit, since the
second branch must always get indented one level farther each time a
commit is squashed. I have some commits where the graph is indented more
than 10 levels due to squashing more than 10 times, making it very
difficult to read.
Reversing the order and printing the second branch before the first
branch prevents this unnecessary indentation and makes the graph easier
to read. This does not change the order of the edges in the graph (i.e.
the first predecessor is still the first edge and the second predecessor
is still the second edge in the graph).
I used to use "remote_branches() & ~mine()" to exclude "their" branches from
the default log, and I don't think that's uncommon requirement. Suppose
untracked branches are usually read-only, it's probably okay to make them
immutable by default.
This helps migrate internal [u8] variables to BStr.
b"" literals in tests are changed to &str to get around potential type
incompatibility between &[u8; N].
Adds support for revset functions `tracked_remote_branches()` and
`untracked_remote_branches()`. I think this would be especially useful
for configuring `immutable_heads()` because rewriting untracked remote
branches usually wouldn't be desirable (since it wouldn't update the
remote branch). It also makes it easy to hide branches that you don't
care about from the log, since you could hide untracked branches and
then only track branches that you care about.
This not only makes the output easier to read, but also protects against
implementation detail changes in `write!` when used with a format
string (especially, how many times and with what strings it calls the
underlying writer).
Partially resolve a 1.5‐year‐old TODO comment.
Add opt‐in syntax for case‐insensitive matching, suffixing the
pattern kind with `-i`. Not every context supports case‐insensitive
patterns (e.g. Git branch fetch settings). It may make sense to make
this the default in at least some contexts (e.g. the commit signature
and description revsets), but it would require some thought to avoid
more confusing context‐sensitivity.
Make `mine()` match case‐insensitively unconditionally, since email
addresses are conventionally case‐insensitive and it doesn’t take
a pattern anyway.
This currently only handles ASCII case folding, due to the complexities
of case‐insensitive Unicode comparison and the `glob` crate’s lack
of support for it. This is unlikely to matter for email addresses,
which very rarely contain non‐ASCII characters, but is unfortunate
for names and descriptions. However, the current matching behaviour is
already seriously deficient for non‐ASCII text due to the lack of any
normalization, so this hopefully shouldn’t be a blocker to adding the
interface. An expository comment has been left in the code for anyone
who wants to try and address this (perhaps a future version of myself).
`tx.format_commit_summary()` can be expensive because it needs to build
an IdPrefixContext now, so it's best to avoid formatting instruction
messages unless they are actually required.
Patience diff starts by lining up unique elements (e.g. lines) to find
matching segments of the inputs. After that, it refines the
non-matching segments by repeating the process. Histogram expands on
that by not just considering unique elements but by continuing with
elements of count 2, then 3, etc.
Before this commit, when diffing "a b a b b" against "a b a b a b", we
would match the two "a"s in the first input against the first two "a"s
in the second input. After this patch, we ignore the "a"s because
their counts differ, so we try to align the "b"s instead.
I have had this commit lying around since I wrote the histogram diff
implementation in 1e657c5331. I vaguely remember thinking that the
way I had implemented it (without this commit) was a bit weird, but I
wasn't sure if this commit would be an improvement or not. The bug
report from @chooglen today of a case where we behave differently from
Git is enough to make me think that we make this change after all.
#761
I've run into change ID prefixes being 4-5 characters instead of the
usual 1-2 characters in commands like `jj duplicate` and `jj split`
fairly often, and it seems like this should resolve that.
For the same reason as the previous commit.
Created and moved stats are printed separately because it's unusual to do both
within one "branch set" invocation.
For the same reason as cdc0cc3601. This will help notice problems like wrong
target revision.
The warning for multiple branches is reorganized as a hint for "-r" option,
which I think is the main purpose of this warning. Unlike "squash", we don't
check if an argument can be parsed as a revset because branch name is usually
a valid symbol expression.
The output looks somewhat similar to color-words diffs. Unified diffs are
verbose, but are easier to follow if adjacent lines are added/removed + modified
for example.
Word-level diffing is forcibly enabled. We can also add a config knob (or
!color condition) to turn it off to save CPU time.
I originally considered disabling highlights in block insertion/deletion, but
that wasn't always great. This can be addressed separately as it also applies
to color-words diffs. #3958
Forgetting a workspace removes its working-copy commit, so it makes
sense for it to be abandoned if it is discardable just like editing a
new commit will cause the old commit to be abandoned if it is
discardable.
It's nice to see the result of "branch move", "create", etc., and this is more
important in "branch move" because the source branches can be specified in an
abstracted way. I originally considered printing a list of affected branches,
but it looked rather verbose. Since the destination revision is unique, we can
use commit_summary template instead.
This patch also removes a warning about multiple branches because the branch
names are included in the commit summary. I think the hint message is good
enough to signal possible mistake.
We usually print stats at the end of mutable operation, and I think these
messages are useful even if N = 1. I understand that "Deleted N" (N > 1) is
unusual and the original intent of these messages was to signal possible
mistakes. However, I don't think printing N=1 stats would nullify the original
purpose.
No emptiness check is needed for delete/forget, but names can be empty in
track/untrack because of noop changes.
The last hunk could be truncated instead, but the .peekable() version is easier
to follow. If we truncated lines, we would have to adjust line ranges
accordingly.
show_context_after was set once when DiffHunk::Different received, and was
never turned off. This means DiffHunk::Matching is not supposed to repeat.
Under this condition, we can assume that removed/added lines exist if lines
isn't empty.
At work, a user encountered a panic upon attempting to create a dir at
the line in the diff below, but it turned out to be difficult to debug
because I didn't know what the path was. There already is a mechanism to
add path context in the lib crate; make it available in the cli crate as
well, and use the mechanism to add path context to "workspace add".
It's common to create empty working-copy commits while using jj, and
currently the author timestamp for a commit is only set when it is first
created. If you create an empty commit, then don't work on a repo for a
few days, and then start working on a new feature without abandoning the
working-copy commit, the author timestamp will remain as the time the
commit was created rather than being updated to the time that work began
or finished.
This commit changes the behavior so that discardable commits (empty
commits with no description) by the current user have their author
timestamps reset when they are rewritten, meaning that the author
timestamp will become finalized whenever a commit is given a description
or becomes non-empty.
Now, the command for `jj git remote add` is `cmd_git_remote_add` and its
argument type is `GitRemoteAddArgs`. This should make it easier to find
the CLI docs and the implementation for commands.
This is how `jj branch` commands were already set up in this way. The
`jj op` commands were also already set up in this way, except the
functions are called e.g. `cmd_op_undo`, I kept this for simplicity.
I was mainly motivated by the `jj file` commands. Most other commands
had functions already named in the above pattern, but used to have
shorter argument type names.
We have two other kinds of working copies at Google and it's sometimes
useful to get the basic information about operation id and tree id for
them, for exampel for debugging stale workspaces. This patch adds a
command for that.
We could instead have made the old `jj debug working-copy` command
work for all kinds of working copies (like the new command) and only
have extra information for the standard local-disk implementation. I
don't feel strongly either way and could do it other other way instead
if people prefer that.
While explaining branch tracking behavior, I find it's bad UX that a deleted
branch can be re-"create"d with tracking state preserved. It's rather a "set"
operation. Since deleted tracking branch is still listed, I think it's better
to assume that the local branch name is reserved.
https://github.com/martinvonz/jj/discussions/3871
Renaming to deleted tracking branch is still allowed (with warning) because the
"rename" command can't handle tracked remotes very well. If it were banned, bad
rename couldn't be reverted by using "jj branch rename". It would be confusing
if "rename a b" succeeded with warning, but the following "rename b a" failed.
This will help inline view.remove_branch() in cmd_branch_forget(). I don't
care much about owned (String, _) vs (&str, _), but we can't simplify the
lifetime issue in find_forgettable_branches() anyway. So I made all callers
pass cloned Arc<ReadonlyRepo> and borrow (name, target) pairs from there.
It was convenient that expression nodes can be compared in tests, but no
equivalence property is needed at runtime. Let's remove Eq/PartialEq to
simplify the extension support.
We now have two `cmd_show` in the repo. I think this one should become
`cmd_file_show`, but this should be done uniformly over all the commands
for consistency.
I did *not* keep `print` as an alias (I couldn't find a compelling
reason to do it), but let me know if anyone feels like keeping it.
We don't want e.g. `jj diff --git` to have underlined text because
it's redundant there. This patch fixes that by adding a new `token`
label used only in the color-words diff (for now - it may be used in
git diffs in the future).
This means we could remove the `line_number` label but I left it
because there's little harm in having it and it seems like it can
still be useful.
Thanks to @yuja for noticing and suggesting the fix.
If you want to set a background color on added/removed lines, you
currently get the same style on the line numbers. This patch lets you
specify a different style by overriding it on the line numbers.
