This patch introduces a new `IndexStore` struct. The idea is that it
will know about the directory in which the index files are stored, the
associations with operations. It may also cache `Arc<ReadonlyIndex>`
instances so if multiple `ReadonlyIndex` instances are loaded, they
can be returned from the cache. That may be useful when merging
operations because the operations are likely to share a large parent
index file. For now, however, all the new type has is `init()`,
`load()`, and `reinit()`.
The only way to load the repo at a current operation (as with
`--at-op`) is currently to first load it at the head operation and
then call `reload()` on the repo. This patch makes it so we can load
the repo directly at the requested operation.
We'll want to be able to load the repo at a given operation without
first loading the head operation as we do today. This patch introduces
a struct for keeping the state of a half-loaded repo. In that
half-loaded state, the store and the op-store have been loaded, but
the view has not yet been loaded. That makes it possible for callers
to use the loaded op-store for looking up an operation to load the
view at.
We want to support loading the repo at a specific operation without
first loading the head operation (like we currently do). One reason
for that is of course efficiency. A possibly more important reason is
that the head operation may be conflicted, depending on how we decide
to deal with operation-level conflicts. In order to do that, it makes
sense to move the creation of the `OpStore` outside of the
`View`. That also suggests that the `.jj/view/op_store/` directory
should move to `.jj/op_store/`, so this patch also does that. That's
consistent with how `.jj/store/` is outside of `.jj/working_copy/`.
All the information needed for calculating the evolution state is now
in the index, so let's use it. This speeds up calculation of the
evolution state from 1.53s to 150ms in the git.git repo. In the Linux
repo, it was sped up from 28.9s to 3.07s. That's still unbearably slow
(and still pretty slow in the git.git repo too). We may need to keep a
persistent cache of the evolution state, but that will have to come
later; this improvement is good enough for now.
Evolution needs to have fast access to the predecessors. This change
adds that information to the commit index.
Evolution also needs fast access to the change id and the bit saying
whether a commit is pruned. We'll add those soon.
Some tests changed because they previously added commits with
predecessors that were not indexed, which is no longer allowed from
this change. (We'll probably eventually want to allow that again, so
that the user can prune predecessors they no longer care about from
the repo.)
The index is now always kept up to date and it has functionality for
finding common ancestors, so let's use it! This should make merging
commits a little faster if their common ancestor is far away (which is
rare). It's probably much more important that the index-based
algorithm is more correct. Also, it returns multiple common ancestors
in the criss-cross case, which lets us do a recursive merge like git
does. I'm leaving the recursive merge for later, though.
We currently need to read the commit objects for finding common
ancestors. That can be very slow when the common ancestor is far back
in history. This patch adds a function for finding common ancestors
using the index instead.
Unlike the current algorithm, which only returns one common ancestor,
the new index-based one correctly handles criss-cross merges.
Here are some timings for finding the common ancestors in the git.git
repo:
| Without index | With Index |
| First run | Subsequent | First run | Subsequent |
v2.30.0-rc0 v2.30.0-rc1 | 5.68 ms | 5.94 us | 40.3 us | 4.77 us |
v2.25.4 v2.26.1 | 1.75 ms | 1.42 us | 13.8 ms | 4.29 ms |
v1.0.0 v2.0.0 | 492 ms | 2.79 ms | 23.4 ms | 6.41 ms |
Finding ancestors of v2.25.4 and v2.26.1 got much slower because the
new algorithm finds all common ancestors. Therefore, it also finds
v2.24.2, v2.23.2, v2.22.3, v2.21.2, v2.20.3, v2.19.4, v2.18.3, and
v2.17.4, which it then filters out because they're all ancestors of
v2.25.3.
Also note that the result was incorrect before, because the old
algorithm would return as soon as it had found a common ancestor, even
if it's not the latest common ancestor. For example, for the common
ancestor between v1.0.0 and v2.0.0, it returned an ancestor of v1.0.0
because it happened to get there by following some side branch that
led there more quickly.
The only place we currently need to find the common ancestor is when
merging trees, which we only do when the user runs `jj merge`, as well
as when operating on existing merge commits (e.g. to diff or rebase
them). That means that this change won't be very noticeable. However,
it's something we clearly want to do sooner or later, so we might as
well get it done.
It's nice to have a non-random order for tests (we can revisit later
if it shows up in profiling). I'm changing the order to be the index
order so the future caller of `heads_pos()` (not `heads()`) will also
get consistent order.
We currently write a new incremental index file every time. That means
that the stack of index files quickly gets deep, which makes it slow
to read the index. This commit makes it so that we squash the new
index segment into its parent if the parent has fewer commits. That
means we'll limit the number of files to O(log n). Writes time will
also be O(log n) on average.
I've confused myself a few times already thinking that level 0 is the
root, so that's probably more intuitive. It also makes tests simpler
because the initial part of the list is unchanged when a new
transaction commits.
With this change, we start writing the incremental index to disk, so
the next reader won't have to re-read the commits and create the
index.
As of this change, we simply write a new index file for each
transaction. That will clearly mean that the stack of files gets deep
pretty quickly. For now, the user will have to do `jj debug reindex`
when things get slow. I plan to change it so instead of writing an
incremental index file every time, we first check if the new index
file would have at least as many commits as the parent file, and if it
will, we write a combined one instead. That should apply recursively,
so we'd have O(log n) index files.
The check for adding an existing commit to the index only checked if
the commit was already in the `MutableIndex`, not if it was already in
the parent `ReadonlyIndex`.
I don't know why I made the walk stop at heads instead of indexed
commits before. Perhaps I did it because it's cheap to check in the
set of head. However, it gets very expensive to walk all the way back
to the root if the parents are not in the set of heads.
With tons of groundwork done, wee can now finally keep the index up to
date within a transaction! That means that we can start relying on the
index to always be valid, so we can use it e.g. for finding common
ancestors within a transaction. That should help speed up `jj evolve`
immensely on large repos.
We still don't write the updated index to disk when the transaction
closes. That will come later.
`Transaction::add_head()` currently invalidates the whole evolution
state. We've had support for incrementally updating evolution since
4619942a57. We should start taking advantage of that. Let's add a
fast-path in `Transaction::add_head()` for the common case where we
add a single commit on top of an existing head. That cheap an simple
to check for. However, it won't cover the case of adding a child off
of a non-head. It's still a good start.
I'm about to move the functions from `CompositeIndex` to an new
`Index` trait implemeted by `ReadonlyIndex` and `MutableIndex`. Since
those types already implement `IndexSegment`, the names would conflict
and it would get annoying to have to disambiguate them. This commit
therefore prepares for that by adding a `segment_` prefix to the
functions in `IndexSegment`.
I want to keep the index updated within the transaction. I tried doing
that by adding a `trait Index`, implemented by `ReadonlyIndex` and
`MutableIndex`. However, `ReadonlyRepo::index` is of type
`Mutex<Option<Arc<IndexFile>>>` (because it is lazily initialized),
and we cannot get a `&dyn Index` that lives long enough to be returned
from a `Repo::index()` from that. It seems the best solution is to
instead create an `Index` enum (instead of a trait), with one readonly
and one mutable variant. This commit starts the migration to that
design by replacing the `Repo` trait by an enum. I never intended for
there there to be more implementations of `Repo` than `ReadonlyRepo`
and `MutableRepo` anyway.
I just learned that attaching a git note is not enough to keep a
commit from being GC'd. I had read `git help gc` before but it was
quite misleading (I just sent a patch to clarify it). Since the git
note is not enough, we need to create some other reference. This patch
makes it so we write refs in `refs/jj/keep/` for every commit we
create. We will probably want to remove unnecessary refs (ancestors of
commits pointed to by other refs) once we have a `jj gc` command.
We store conflicts as blobs with JSON data and with a git note
pointing to them to prevent GC. These are stored in the git tree as
regular files. The only thing that distinguishes them is that their
filename ends with `.jjconflict`. Since they are referenced from the
tree, there's no need for the git note to prevent GC (which doesn't
work anyway, as I just learned), and we don't store any additional
data in the note either, so let's just remove it.
Windows doesn't support recording the executable bit in the file
system. Before this commit, the code for reading and writing the
executable wouldn't even compile on Windows. This commit at least
makes it so we preserve whatever bit has been recorded in the repo.
At least I hope that's what it does -- I don't have access to a
Windows machine right now.
The project doesn't currently build on Windows. One reason is because
we had a `unimplemented!()` when trying to write a symlink. Let's
print a warning instead, so the project can start building on
Windows. (The next patch will fix another build problem on Windows.)
When a transaction gets dropped without being committed or explicitly
discarded, we currently raise an assertion error. I added that check
because I kept forgetting to commit transactions. However, it's quite
normal to want to drop transactions in error cases. The current
assertion means that we panic and don't report the actual error to the
user in such cases. We should probably audit the code paths where we
commit transactions and decide for each if we simply want to to
discard the transaction or not. In some cases, we may want to commit
the transaction without integrating it in the operation log
(i.e. without creating a file entry in .jj/views/op_heads/). However,
we can do that later. For now, let's just make sure we don't panic
when dropping the transaction in release builds.
I'm about to move `MutableRepo` to the `repo` module so it becomes
more important to encapsulate access. Besides, the new functions
introduced in this commit reduces some duplication.
There's still one access of `MutableRepo::evolution` in
`Transaction::new()`. I'll address that next by adding a factory
function to `MutableRepo`.
I've been confused twice that rebasing an open commit so it results in
conflicts doesn't show the conflicts in the log output. That's because
we create a successor instead if a commit with conflicts is open. I
guess I thought it would be expected that a child commit was not
created. Since it seems surprising in practice, let's change it and
we'll see if the new behavior is more or less surprising.
The most annoying remaining bug is that 3-way merge frequently panics
with "unhandled merge case". This commit fixes that by rewriting the
merge code. The new code is based on the algorithm used in Mercurial
(which was in turn copied from Bazaar):
1. Find "sync" regions, which are regions that are the unchanged in
the base and two sides. Note their start end end positions in each
version.
2. Produce the output by taking the sync regions and inserting the
result of merging the regions between the sync regions. These
regions can either be changed on only one side, in which case we
use that version, or it can be changed on both sides, in which
case we indicate a conflict in the output.
It's both more correct and much easier to follow.
We want to be able to be able to do fast `.contains()` checks on the
result, so `Iterator` was a bad type. We probably should hide the
exact type (currently `HashSet` for both readonly and mutable views),
but we can do that later. I actually thought I'd want to use
`.contains()` for indiciting public-phase commits in the log output,
but of course want to also indicate ancestors as public. This still
seem like a step (mostly) in the right direction.
Mercurial's "phase" concept is important for evolution, and it's also
useful for filtering out uninteresting commits from log
output. Commits are typically marked "public" when they are pushed to
a remote. The CLI prevents public commits from being rewritten. Public
commits cannot be obsolete (even if they have a successor, they won't
be considered obsolete like non-public commits would).
This commits just makes space for tracking the public heads in the
View.
All commits in the view are supposed to be reachable from its
heads. If a head is removed and there are git refs pointing to
ancestors of it (or to the removed head itself), we should make that
ancestor a head.
The only invariant we currently enforce is that the set of heads does
not include any ancestors of other commits in the set. I'm about to
make sure that we don't end up with dangling git refs (pointing to
commits no reachable from the heads). It will be useful to have a
single place to enforce that since we'll need to do the same thing
after updating the view as after merging views.
I think it's better to let the caller decide if the parents should be
added. One use case for removing a head is when fetching from a Git
remote where a branch has been rewritten. In that case, it's probably
the best user experience to remove the old head. With the current
semantics of `View::remove_head()`, we would need to walk up the graph
to find a commit that's an ancestor and for each commit we remove as
head, its parents get temporarily added as heads. It's much easier for
callers that want to add the parents as heads to do that.
Git refs are important at least for understanding where the remote
branches are. This commit adds support for tracking them in the view
and makes `git::import_refs()` update them.
When merging views (either because of concurrent operations or when
undoing an earlier operation), there can be conflicts between git ref
changes. I ignored that for now and let the later operation win. That
will probably be good enough for a while. It's not hard to detect the
conflicts, but I haven't yet decided how to handle them. I'm leaning
towards representing the conflicting refs in the view just like how we
represent conflicting files in the tree.
This is partly to prepare for merging the operations in order of
transaction-commit time (currently merged in order of operation id),
so we can get a predictable order in tests (assuming transactions are
not committed the same millisecond).
The function used to be larger when we had more reference cycles
between e.g. the `WorkingCopy` and `Repo`. Now there's only
`Evolution` left, so let's inline the function.
This commits makes it so that running commands outside a repo results
in an error message instead of a panic.
We still don't look for a `.jj/` directory in ancestors of the current
directory.
I'm preparing to publish an early version before someone takes the
name(s) on crates.io. "jj" has been taken by a seemingly useless
project, but "jujube" and "jujube-lib" are still available, so let's
use those.
I had missed in `git2-rs`'s documentation that you need to check
in a callback if the remote ref(s) got updated by the push or
not. This adds such a check and a new error variant for rejected
branch updates.
I tried to push a commit from my Jujube repo to GitHub using `jj
git push --branch main` and it became clear that we need to pass
SSH credentials. This commit hopefully fixes that. I've only made
it pass credentials for ssh-agent for now, because that seems to
be enough to make it work for me personally. If this commit
becomes visible on GitHub, it should mean that it worked.
When you run e.g. `jj st` outside of a repo, it just
crashes. That'll probably give new users a bad impression, so I
was planning to improve error handling a bit. A good place to
start is by fixing the code I recently added (which obviously
should have been using `thiserror` from the beginning). That's
what this commit does.
Also, this is the first commit in this repo created with
Jujube! I've just started dogfooding it myself.
I've forgotten to close a transaction a few times and while the
message ('assertion failed: self.closed') is clear to me now, it
probably won't be clear to others or to me in the future.
This adds `jj git fetch` for fetching from a git remote. There remote
has to be added in the underlying git repo if it doesn't already
exist. I think command will still be useful on typical small projects
with just a single remote on GitHub. With this and the `jj git push` I
added recently, I think I have enough for my most of my own
interaction with GitHub.
The fact that no commits from the underlying Git repo were imported
when creating a new Jujube repo from it was quite surprising. This
commit finally fixes that.
When using Git as a store, new commits created in the underlying Git
repo are only made visible by making changes on top of them (e.g by
checking them out, so a working copy commit is created on top). That's
especially confusing when creating a new repo backed by an existing
Git repo, because the commits from that repo don't show up.
This commit prepares for fixing that by adding a function for updating
heads based on git refs. Since we don't yet track git refs (or
anything similar), the function just makes sure the refs are visible
in the Jujube repo by making them (anonymous) heads.
`Transaction::add_head()` and others would let the caller add
non-heads to the set (i.e. ancestors of others heads) and the the
non-heads were filterd out when the transaction was committed. That's
a little surprising, so let's try to keep the set valid even within a
transaction. That will surely make commands that add many commits
noticeably slower in large repos. Hopefully we can improve that
later.
Returning the store's internal `git2::Repository` instance wrapped in
a `Mutex` makes it easy to run into deadlocks. Let's return a freshly
loaded repo instance instead.
It's annoying to have to have the Git repo and Jujube repo in separate
directories. This commit adds `jj init --git`, which creates a new
Jujube repo with an empty, bare git repo in `.jj/git/`. Hopefully the
`jj git` subcommands will eventually provide enough functionality for
working with the Git repo that the user won't have to use Git commands
directly. If they still do, they can run them from inside `.jj/git/`,
or create a new worktree based on that bare repo.
The implementation is quite straight-forward. One thing to note is
that I made `.jj/store` support relative paths to the Git repo. That's
mostly so the Jujube repo can be moved around freely.
This commit starts adding support for working with a Jujube repo's
underlyng Git repo (if there is one). It does so by adding a command
for pushing from the Git repo to a remote, so you can work with your
anonymous branches in Jujube and push to a remote Git repo without
having to switch repos and copy commit hashes.
For example, `jj git push origin main` will push to the "main" branch
on the remote called "origin". The remote name (such as "origin") is
resolved in that repo. Unlike most commands, it defaults to pushing
the working copy's parent, since it is probably a mistake to push a
working copy commit to a Git repo.
I plan to add more `jj git` subcommands later. There will probably be
at least a command (or several?) for making the Git repo's refs
available in the Jujube repo.
This adds an interactive mode for `jj restore`. It works by first
creating two temporary directories with the contents of the subset of
files that differ between the two trees, and then letting the user
edit the directory representing the right/after side. This has some
advantages compared to the interactive modes in Git and Mercurial:
* It lets the user edit the final state as opposed to the diff itself
(depending on the diff tool, of course). I think most users find it
easier to edit the file contents than to edit the patch
format.
* It delegates the hard work to a tool that is already written (this
is a big advantage for an immature tool like Jujube, but it is not
an advantage from the user's point of view).
Almost all of the work in this commit went into adding a function that
takes two trees, lets the user edit the diff, and returns a new tree
id. I plan to reuse that function for other interactive commands. One
planned command is `jj edit`, which will let the user edit the changes
in a commit. `jj edit -r abc123` will be mostly about providing a more
intuitive name for `jj restore --source abc123^ --destination abc123`,
plus it will be different for merge commits (it will edit only the
changes in the merge commit). I also plan to add `jj split` by letting
the user edit the full diff, leaving only the parts that should go
into the first commit. Perhaps there will also be commands for moving
part of a commit out of or into a parent commit.
This extracts a bit from `Transaction::check_out()` for taking a
Conflict, materializing it, and writing the resulting plain file to
the store. It will soon be reused.
Before this commit, running Git's GC in a Git repo backing a Jujube
repo would risk deleting the conflict data we store as blobs in the
Git repo. This commit fixes that by adding a Git note pointing to the
conflict blob.
I wasn't able to add a test case for this because libgit2 doesn't
support gc [1]. Just testing that the ref is there doesn't seem very
useful.
[1] https://github.com/libgit2/libgit2/issues/3247
Very little code was saved by reusing `write_file()` and it made it
confusing (needed to provide an unused filename). Also, I'll soon want
access to the `locked_repo` variable in `write_conflict()`.
I have never run into this being a problem in practice, but this
change is a stepping stone for two things:
1. Using exponential backoff for other locks (in particular the
working copy).
2. Making the Git store write a ref for conflict objects, so they
don't get GC'd (I want to do that before even I start dogfooding).
I had tried to generate the protobuf code at build time many months
ago, but decided against it because it slowed down the build too
much. I didn't realize there was the
"cargo:rerun-if-changed=<filename>" feature that time. Given that that
exists, it seems like an obvious win to generate the source code at
build time.
I put the generated sources in `$OUT_DIR` (where [1] says they should
be), then include them in the `protos` module by using the `include!`
macro. The biggest problem with that is that I couldn't get IntelliJ
to understand it, even after enabling the experimental features
described in [2].
[1] https://doc.rust-lang.org/cargo/reference/build-script-examples.html#code-generation
[2] https://github.com/intellij-rust/intellij-rust/issues/1908#issuecomment-592773865
We currently recalculate the entire evolution state whenever a new
commit is added within a transaction. That's clearly wasteful. This
commit makes the state-update incremental.
I'm about to make the evolution state updated incrementally. To be
able to tell if a new commit is divergent, we'll need to keep track of
already existing, non-obsolete commits in the change, even if they're
not divergent before the new commit is added.
A pruned commit just indicates that its predecessors should be evolved
onto the pruned commit's parent instead of onto the pruned commit
itself. The pruned commit itself can be divergent. For example, if
there are several pruned sucessors of a commit, then it's unclear
where the predecessor's children should be rebased to.
I'm about to add more fields to the type and this will help to
slightly reduce the boilerplate for initializing and the maps and sets
and creating the struct.
Before this commit, when the `evolve()` evolved a stack of orphans, it
would use the evolve state from the beginning of the function to
calculate where they should go. That meant that only the bottom-most
orphan(s) would get evolved to their right place. This commit fixes
that by use the Transaction's evolution state.
Before this commit, it could share its state with the
`ReadonlyEvolution`. That makes no sense when the state is modified,
and would crash if we tried to get a mutable reference to the
state. It only "worked" because the state is not yet updated within a
transaction (a known bug). I'm about to fix that bug, so we need to
fix the ownership, which this commit does.
There was some duplicate between `ReadonlyEvolution` and
`MutableEvolution` that could be extracted. It will help to have this
shared code on the `State` object for the next few patches.
The point of having the `modified` and `removed` files in the test was
to show that they don't get untracked, but I forgot to include them in
the `.gitignores`, so there was no reason they would have gotten
untracked anyway.
The project's source of truth is now in Git and I really miss support
for anonymous heads and evolution (compared to when the code was in
Mercurial). I'm therefore more motivated to make the tool useful for
day-to-day work on small repos, so I can use it myself. Until now, I
had been more focused on improving performance when it was used as a
read-only client for medium-to-large repos.
One important feature for my day-to-day work is support for
ignores. This commit adds simple and effective, but somewhat hacky
support for that. libgit2 requires a repo to check if a file should be
ignored (presumably so it can respect `.git/info/excludes`). To work
around that, we create a temporary git repo in `/tmp/` whenever the
working copy is committed. We set that temporary git repo's working
copy to be shared with our own working copy. Due to
https://github.com/libgit2/libgit2sharp/issues/1716 (which seems to
apply to the non-.NET version as well), this workaround unfortunately
leaves a .git file (pointing to the deleted temporary git repo) around
in every Jujube repo. That's always ignored by libgit2, so it's not
much of a problem.
This removes one level of indirection, which is nice because it was
visible to the callers. The `Index` struct is now empty. The next step
is obviously to delete it (and perhaps rename `IndexFile` to `Index`
or `ReadonlyIndex`).
