The semantics is similar to experimental.directaccess=true in Mercurial. Hidden
revisions and their ancestors become temporarily available. This means all() is
not exactly the same as ::visible_heads(). The latter never includes hidden
revisions.
We could instead transform all() to (all() | referenced_commits). However, this
wouldn't work if expressions like ::hidden are intersected/united with filters,
all(), etc.
Fixes#5871
This was discussed in #3219 and #3262, and apparently, people don't agree
whether we should implement machine-readable formatting by using templater.
However, there are recurring requests about JSON output, and "jj api" is still
vaporware. So I think it's good to add some support for JSON output.
I'm not going to add syntax to build custom JSON object/list for now. Instead,
property types such as Commit will implement fixed serialization output. The
user might have to write ad-hoc JSON constructs to concatenate the default JSON
output and their custom JSON fields.
We'll need a documentation about which types are serializable. I think this can
be added later when more types get serialization support.
Currently, this is parsed as `x..visible_heads()`, which prevents
optimizing `x.. & ::y` to `x..y`.
An alternative would be to remove `::visible_heads()` from
intersections, similar to `all()`, but it might be better to avoid
removing `visible_heads()` if it was explictly requested by a user.
- Use `jq -s` to read the entire input (could be several arrays due to GitHub API
pagination) before running the filter just once.
- Deduplicate by `.author.login` in case a person with the same GitHub username has
commits under different names within a release.
- Use string interpolation to format each line.
This version describes a design that I think will work relatively well
for supporting copy tracing even when there are conflicts.
There are still some open questions, in particular around how to
populate the data in Git repos, and how to efficiently query the data
in very large repos like the one at Google. I think it's still clear
enough that we can start working on the implementation.
The commit id prefix here should be `f7fb5943`, from the working copy's parent before
the squash (on line 351). Also see line 413 where the full commit id is shown in the
`Undid operation` message.
The optional arguments for each shell were deprecated in #2945 and v0.19.0.
The `shell` argument is now required, and Bash completions are no longer generated by
default if `jj util completion` is run without any arguments.
It isn't necessary to include shebangs in scripts passed into
`util exec` since you're already explicitly calling out the shell you'd
like to use. This change removes shebangs from examples for `util exec`
and adds a note below calling this out.
This allows users to override the external tool set in ~/.jjconfig.toml. The
config variable is renamed to ui.diff-formatter to be consistent with the other
diff/merge editor settings.
--tool=:<format> support will be added separately.
Closes#3327
Since jj operates in detached HEAD state, a tracking bookmark isn't technically
needed in order to check out a commit. I usually employ this and delete
unmanaged local bookmarks after cloning repository. This option will automate
the setup.
I think its time to finally give our Conflict Algebra rule a term.
This is will make our documentation alot easier, and gives it a name for
new design docs.
All credit goes to @martinvonz for coming up with the term in #4988.
This is the last part of the series 02722eae54 "view: port bookmark/tag name
types to RefName/RemoteName." Templater doesn't use these name types internally
because we wouldn't gain much type safety in template language.
The return type of .remote() is changed to Option<RefSymbol>. An empty symbol is
rendered with quotes, so it would be weird if local_bookmark.remote() returned
an empty RefSymbol.
I'm going to add RefSymbol(String) type to format bookmark/remote name in revset
syntax. Since an empty symbol will be rendered differently from an empty string,
bookmark.remote() -> String will be changed to -> Option<RefSymbol>. Then, we'll
need to add support for comparison between optional and non-optional values.
In template language, method call on None produces an error. If we followed this
rule, comparison would have to be expressed as `opt && opt == lit`. I feel this
would be too verbose, and wouldn't help eliminate user error caused by
unintended comparison between Option<T> and T anyway. So this patch implements
comparison by lifting T to Option<T> implicitly.
I spent quite a while confused by the presence of
`if(!trailers.contains_key("Change-Id")` in the `commit_trailers` example. My
thinking was: if `commit_trailers` already performs deduplication, then isn't
this logic useless?
Eventually I realized that the `commit_trailers` deduplication works on the
entire line, not just the key as I'd assumed. (And that makes sense for
trailers like Signed-off-by that can appear more than once.)
I've reworded the `commit_trailers` documentation to try to make this more
clear.
`heads(tags())` was rather confusing in my repository because I have
a bunch of tags pushed to my fork that are tracked locally. Assuming
that `trunk()` is something like `main@origin` or whatnot, then using
`heads(tags() & ::trunk())` works quite a bit better.
Signed-off-by: Austin Seipp <aseipp@pobox.com>
