git-repository diff API #535

Byron · 2022-09-19T09:46:37Z

Now that it becomes clearer what to expect from diffing, we can go ahead and provide a high-level
API that can be used by crates-index-diff and by ein tool hours as well.

It should provide basic integration with git-diff, which for now is probably now more than a
type-based wrapper for similar.

Tasks

High-level API in git-repository
ability to diff two files via git-diff and Myers algorithm
upgrade crates-index-diff to that version of git-repository, for now in a branch
add ein tool hours --stats to get line-stats per author
percentages for lines and files
new release for crates-index-diff to use and new release of the latter.
basic tests for new functionality
make path component iteration easier for lookup_path().

Out of scope

rename tracking

…unction (#470)

…#470) Even though the empty-tree object can be found when searched via `Repository::find_object()`, previously it was not locatable when used during rev-spec parsing.

…e. (#470) This is a first and maybe the last step towards providing diffing functionality, and it seems like the right choice to keep this in similar and contribute there as needed. All algorithms are well described and thus shouldn't be git-specific per-se, and `similar` is the best the community has to offer.

…line. (#470)

…mmit ancestor iteration. (#470) This is useful to avoid additional lookups of the same object for reading additional data from it. Currently one needs an object cache to avoid duplciate object extraction work, with such a cache being slower than accessing the same buffer again.

…ut consumption. (#470) This is useful if these provide additional out-of-band information.

…thor. (#470) Note that these are expensive and unconditionally use threads to speed up these computations.

This allows path conversions there to be just as efficient as on unix. This was adopted from [a PR in the hexlix-editor](https://github.com/helix-editor/helix/pull/3890/files#diff-504515b66023120e75a921cd56a932aed76c0ff62593fbb69d92e0ef65089501R47).

The disadvantage of doing that is that we copy the name, which allocates, instead of just returning it by reference.

It's very common to try to work with trees, so let's make that easier.

But for some reason, the counter stays at 0 despite sharing the counter stat.

Finally, issues due to bugs in prodash.

…modified. (#470)

Even though failing is cheap, it's not free and can be done a million times.

…line-stats`. (#470) That way more information is generated at increasingly high costs.

…dency graph (#470)

…). (#470) We assume unguessable ones are a plain text file, like one without extension.

…actor v0.12.0, git-glob v0.4.0, git-path v0.5.0, git-quote v0.3.0, git-attributes v0.4.0, git-config-value v0.8.0, git-tempfile v2.0.5, git-validate v0.6.0, git-object v0.21.0, git-ref v0.16.0, git-sec v0.4.0, git-config v0.8.0, git-discover v0.5.0, git-traverse v0.17.0, git-index v0.5.0, git-worktree v0.5.0, git-testtools v0.9.0, git-command v0.1.0, git-prompt v0.1.0, git-url v0.9.0, git-credentials v0.5.0, git-diff v0.19.0, git-mailmap v0.4.0, git-chunk v0.3.2, git-pack v0.23.0, git-odb v0.33.0, git-packetline v0.13.0, git-transport v0.20.0, git-protocol v0.20.0, git-revision v0.5.0, git-refspec v0.2.0, git-repository v0.24.0, git-commitgraph v0.9.0, gitoxide-core v0.18.0, gitoxide v0.16.0, safety bump 28 crates SAFETY BUMP: git-actor v0.12.0, git-attributes v0.4.0, git-config-value v0.8.0, git-object v0.21.0, git-ref v0.16.0, git-sec v0.4.0, git-config v0.8.0, git-discover v0.5.0, git-traverse v0.17.0, git-index v0.5.0, git-worktree v0.5.0, git-testtools v0.9.0, git-command v0.1.0, git-prompt v0.1.0, git-url v0.9.0, git-credentials v0.5.0, git-diff v0.19.0, git-mailmap v0.4.0, git-pack v0.23.0, git-odb v0.33.0, git-transport v0.20.0, git-protocol v0.20.0, git-revision v0.5.0, git-refspec v0.2.0, git-repository v0.24.0, gitoxide-core v0.18.0, gitoxide v0.16.0, git-pathspec v0.1.0

…edentials v0.5.0, git-diff v0.19.0, git-mailmap v0.4.0, git-chunk v0.3.2, git-pack v0.23.0, git-odb v0.33.0, git-packetline v0.13.0, git-transport v0.20.0, git-protocol v0.20.0, git-revision v0.5.0, git-refspec v0.2.0, git-repository v0.24.0, git-commitgraph v0.9.0, gitoxide-core v0.18.0, gitoxide v0.16.0

Byron added 2 commits September 19, 2022 17:38

refactor (#470)

6ac7dbe

refactor (#470)

a938fe4

Byron mentioned this pull request Sep 19, 2022

gix towards 1.0 #470

Open

60 tasks

Byron added 27 commits September 19, 2022 17:59

diff platform for basic diff configuration (#470)

c857b9b

break through API surface and sketch delegate calling user-provided f…

e51f3cd

…unction (#470)

allow user callbacks to have any error (#470)

5be96b3

first test for simple file modification detection (#470)

a9056fd

Provisions for tracking the location of a change. (#470)

7fd9b0e

Support for file-name tracking (#470)

88c4a57

improved usability of the Action enum (#470)

d04807b

fix docs (#470)

9b7aaa0

refactor (#470)

90b9c90

refactor (#470)

9d01fb4

fix: rev-spec parsing can now handle the empty tree as full hex hash. (…

ae38660

…#470) Even though the empty-tree object can be found when searched via `Repository::find_object()`, previously it was not locatable when used during rev-spec parsing.

Support for Path tracking (#470)

64bbb3d

performance note (#470)

0670468

Slightly improved docs for traversal docs. (#470)

963055b

feat: Once a change is obtained, it's easy to obtain changes line by …

8c2e5c6

…line. (#470)

feat: interrupt::Iter now allows accessing the inner iterator witho…

1027be9

…ut consumption. (#470) This is useful if these provide additional out-of-band information.

feat: ein tool hours --stat to collect additional statistics per au…

28c4cae

…thor. (#470) Note that these are expensive and unconditionally use threads to speed up these computations.

a little more complexity for diff tests (#470)

5878ad1

simplify looking up entries by path (#470)

15a18e4

rename!: Tree::lookup_path() -> Tree::lookup_entry(). (#470)

79c2255

adapt to changes in git-repository (#470)

5ec714f

Add note on why we consume the tree for looking up an entry. (#470)

b285097

The disadvantage of doing that is that we copy the name, which allocates, instead of just returning it by reference.

thanks clippy

77ff8ae

fix docs (#470)

593f57b

Byron added 19 commits September 20, 2022 09:45

upgrade to prodash 20.1 for Progress::counter() feature (#470)

0ac4a2c

feat: Object::peel_to_tree() as convenience method. (#470)

0871a96

It's very common to try to work with trees, so let's make that easier.

First attempt to get progress information from stat worker. (#470)

0947c70

But for some reason, the counter stays at 0 despite sharing the counter stat.

working progress printing (#470)

67ec2c7

Finally, issues due to bugs in prodash.

feat: ein tool hours -s shows statistics about files added/removed/…

b8f2f8b

…modified. (#470)

avoid binary search if there is nothing to find (#470)

ffd4f0f

Even though failing is cheap, it's not free and can be done a million times.

change: ein tool hours -s was split into -f|--file-stats and `-l|…

3c7c9a7

…line-stats`. (#470) That way more information is generated at increasingly high costs.

show discovered file chnages in real-time (#470)

0614318

collect changes lines as well (#470)

84eb777

make stat extractions interruptable (#470)

cb53324

refactor (#470)

e3e3211

replace flume with crossbeam-channel as it's already in the depen…

e27b65a

…dency graph (#470)

only diff files that are probably not a binary file (as per extension…

2a3c5b0

…). (#470) We assume unguessable ones are a plain text file, like one without extension.

percentages for files and lines added (#470)

bf63a13

Some tests to pin the new statistical capabilities (#470)

47e2370

update changelogs prior to release (#470)

caa7a1b

fix dependency declaration (#470)

9a8369d

Byron merged commit f5c36d8 into main Sep 20, 2022

Byron deleted the diff branch September 20, 2022 10:51

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

git-repository diff API #535

git-repository diff API #535

Byron commented Sep 19, 2022 •

edited

Loading

git-repository diff API #535

git-repository diff API #535

Conversation

Byron commented Sep 19, 2022 • edited Loading

Tasks

Out of scope

Byron commented Sep 19, 2022 •

edited

Loading