This helps propagate error from tx.write(). I made the error variants
non-transparent because it's easier than manually implementing From<> that maps
each inner error to the other inner error.
When read/writing commits from the git-backend, populate the git commit
header with a backwards hash of the `change-id`. This should enable
preserving change identity across various git remotes assuming a
cooperative git server that doesn't strip the git header.
This feature is behind a `git.write-change-id-header` configuration flag
at least to start.
The original idea was to flatten left/right conflict trees and pair up adjacent
negative/positive terms. For example, diff(A, B-C+D) could be rendered as
diff(A, B) and diff(C, D). The problem of this formalization is that one of the
diff pairs is often empty (because e.g. A=B), so the context is fully omitted.
The resulting diff(C, D) doesn't provide any notion why the hunk is conflicted,
and how it is different from A.
Instead, this patch implements diffs in which each left/right pair is compared.
In the example above, the left terms are padded, and the diffs are rendered as
diff(A, B), diff(-A, -C), diff(A, D). This appears to be working reasonably well
so long as either side is resolved or both sides have the same numbers of terms.
Closes#4062
This helps extract hunk rendering function for non-materialized color-words
diffs. In conflict hunk, an identical diff pair will be omitted with "..."
marker.
This patch introduces a subtle behavior change when "ignore whitespace" options
are used. Before, "..." wouldn't be printed if contents differ only in
whitespace. I don't think the new behavior is bad because the file header says
"Modified regular file" in that case.
I originally thought we could remove args.reset_author tests in later pass, but
we can't because author timestamp may be updated if a commit is discardable.
Still it's nice that we can deduplicate some logic.
It can be useful for fixers to record some information about the
inputs they processed. For example, the `FileFixer` we use in our
server at Google get more detailed error information back from its
formatter tools. We may want to record these errors in our
`FileFixer`'s state so we can then return that to the client.
The `new_parents_map` will allow the parents of each commit to be
customized before the `transform_descendents` callback is invoked. This
is useful when the order of commits needs to be changed, whereby setting
the new parents in the default callback might lead to repeated rebasing
and cycles if the new parent has not been visited in the reverse
topological order traversal.
These merge tools didn't work properly on conflicted trees with more
than two sides since they didn't simplify the conflict before resolving.
The new implementation is more similar to how external merge tools are
executed, which should give a more consistent behavior.
With this template, a 'git format-patch' compatible
email message can be generated using something like:
jj show --git --template git_format_patch_email_headers <rev>
Since 7618b52b "cli: consider 'JJ:' lines as comments also when not followed by
space", lines starting with "JJ:" (without space) are also ignored. We can
simply add "JJ:" prefix to empty intro/instruction lines.
Closes#5484
This helps detect whether the last line is "JJ:" instruction or not. It seems
also nice that I don't have to insert newline to reflow the edited paragraph.
When rendering diff of conflicts, it might make sense to show diff of negative
(or base) terms with "removed"/"added" labels swapped. For example, diffs
between A and B-C+D can be rendered as
{left, right} = {A (-), B (+)}, {A (+), C (-)}, {A (-), D (+)}
by padding -A+A to the left side. To achieve that, I'm thinking of adding
labels: [&str; 2] parameter. This patch will help keep function arguments more
consistent there.
FWIW, I also tried rendering in {A (-), B (+)}, {C (-), D (+)} forms. This
seemed not intuitive because, when diffing, we compare two distinct states. In
the example above, the diff {C, D} should be considered an internal state at the
right side, not the states pair to be diffed.
I think this makes more sense because WorkspaceId is currently a human-readable
name. In error/status messages, workspace names are now printed in revset
syntax.
New WorkspaceId types do not implement Default. It would be weird if string-like
type had non-empty Default::default(). The DEFAULT constant is provided instead.
It seemed inconsistent that only GitExportError was translated to an internal
error. There are various reasons that triggers git::export_refs()/reset_head()
failure, and some of them are environmental error.
Wrapped errors are usually displayed after "Failed to import refs ...", etc., so
the context is obvious. I also removed "Internal"/"Error" for consistency.
The option will be used to prioritize branches of the `jj log` graph to
be displayed on the left.
This can make them more readable in some situations.
An example would be
```
[revsets]
log-graph-prioritize = "coalesce(description("megamerge\n"), trunk())"
```
It's deprecated since 1aad250420 "Shorten the git push branch when possible
using the short change ID hash" (2023-01-12). I don't think we need the
fallback. I also removed the check for ambiguous prefixes as I believe it
wouldn't practically matter. If needed, we can add a check for existing bookmark
pointing to different commit. We can also make it templated with default
"'push-' ++ change_id.shortest(12)".
