Welcome to the 67th edition of Git Rev News, a digest of all things Git. For our goals, the archives, the way we work, and how to contribute or to subscribe, see the Git Rev News page on git.github.io.
This edition covers what happened during the month of August 2020.
The Google Summer of Code 2020 officially ended on September 8, 2020. Congratulations to the three students who were mentored by the Git community, as they successfully passed all the evaluations! Their final reports and code submissions are available:
Hariom Verma, who was mentored by Heba Waly and Christian Couder, worked on “Unify ref-filter formats with other --pretty formats”.
Abhishek Kumar, who was mentored by Jakub Narębski and D. Stolee, worked on “Implement Generation Number v2”.
Shourya Shukla, who was mentored by Kaartic Sivaraam and Christian Couder, worked on “Convert submodule to builtin”.
[RFC PATCH 0/2]extend --abbrev support to diff-patch format
Đoàn Trần Công Danh sent a small RFC patch series that makes the diff
machinery, which is used by git diff and similar commands, support
the --abbrev option when it shows a patch.
In addition to the familiar --patch format, which is used when
showing regular diffs in a way similar to the
unified format of the Unix diff command,
the diff machinery can produce the --raw format output to show object
names before and after the change for each path like this:
:100644 100644 86e5411f39 f1b1bc8aa0 M Makefile
In the --patch format, object names before and after the change
appear on the index line, which looks like this:
index 4adb9ce124..67d9801da3 100644
The --abbrev option used to control the abbreviation of the
object names in the --raw format but did not affect those shown
on the index line in the --patch format. Đoàn’s patch series
makes the option control both.
In the cover letter of his RFC patch series, Đoàn explained how the changes would improve his workflow when dealing with patches. But Junio Hamano and Jeff King, alias Peff, replied saying that they had trouble understanding the purpose of the series from the explanations in the cover letter.
After a discussion with Peff and Junio, it turns out that using
the --full-index option is a better approach to help Đoàn’s workflow
(the option was specifically designed to ensure that the index
line in the --patch format can be used to exactly identify the
objects involved in the change to help the users, which was what
Đoàn’s workflow wanted). However, a consensus was reached that
Đoàn’s patch series was worth moving forward, because it improved
consistency between the diff formats.
So Đoàn sent a version 2 of his patch series with only changes in the cover letter and in the commit message of the second patch.
Junio replied to the first patch of the series though. This patch
changed how --no-abbrev was dealt with. Junio asked for
clarifications in the commit message, especially about why the patch
was necessary.
So Đoàn sent a version 3 of his patch series with only changes in the commit message of the first patch. After further discussion with Junio, it appears that the patch was not actually necessary.
In the meantime though Gábor Szeder found that the tests in the
second patch were failing when run with
GIT_TEST_DEFAULT_HASH=sha256. This setting makes Git use the
SHA-256 hash algorithm
instead of the usual
SHA-1 hash algorithm
for object names.
A discussion about the best approach to fix this then followed between Đoàn, Junio and Brian M. Carlson, who has been working for a long time on implementing support for SHA-256 in Git.
Đoàn eventually sent a version 4 of his patch series, where the first patch of his version 3 had been discarded, but a new patch to improve handling of SHA-256 object names in the tests replaced it.
Junio and Đoàn discussed further improvements to the commit messages of both patches. Then Đoàn sent a version 5 taking these improvements into account.
The patch series has since been merged into the ‘master’ branch.
Who are you and what do you do?
I’m just another average developer addicted to an exotic environment. Outside of my $dayjobs, I maintain Git and some other packages for VoidLinux.
What would you name your most important contribution to Git?
I would consider fixing remaining problems with musl libc my most important contributions to Git, since I mostly stay inside my musl box these days.
What are you doing on the Git project these days, and why?
I mostly watch for sideline these days, since I’m busy with other things. Occasionally, I jump in one or two discussion since that topic looks relevant to me. Sometimes, I write a patch or two to support my own jobs.
If you could get a team of expert developers to work full time on something in Git for a full year, what would it be?
I don’t have anything specific in mind, except the transition into the new hash algorithm, brian m. carlson has done a very good job, and I understand the transition would require a long transition period, testing, a lot of testing.
If you could remove something from Git without worrying about backwards compatibility, what would it be?
The diff to merge-base notation (upstream…fork) of the diff family. This has been floating in the mailing list for a while, and I won’t waste this opportunity to not mention it again.
There’s an on-going change from Denton Liu to support --merge-base
into the diff family. Hopefully, people can retrain their brain-muscles to
use this new option instead.
Events
Various
Set the default branch for newly-created GitHub repositories
on GitHub Blog.
On October 1, 2020, if you haven’t changed the default branch
for new repositories for your user, organization, or enterprise on
GitHub, it will automatically change from master to
main. You can opt out of this change at any time. Existing
repositories will continue to have the same default branch they have
now.
‘There is no way we can keep coding local’: GitPod’s cloud development platform released into sunlight of open source – The Register chats with co-founder Sven Efftinge.
Discussions have been going on about the Git project participating in the next Outreachy round. So far 3 projects have been proposed.
Light reading
Guillaume Papin in his development notes on a compilation database (a concept that includes, but is not limited to a JSON Compilation Database) includes section on Git as one of (and at this time the only) case studies on open source projects.
There is a patch by Philippe Blain to add support for generating a JSON compilation database to Git’s Makefile. It is currently merged into ‘next’.
Keith Peters writes how to create your own Git-based Wiki to save information that one might need later. The idea is to make use of the fact that many Git hosting sites render Markdown on-line, by storing the data in interlinked Markdown files – the article proposes the template to use.
Patterns for writing better git commit messages
by Helder Burato Berto, with references to other articles, and
a git commit template.
Learn Enough Git and Github to take part in the Weekly Challenge by Dave Cross (not intended as “beginners guide to Git and Github”).
Git tools and sites
ugit: DIY Git in Python is a tutorial to help learn about Git internals by building an implementation of Git in Python.
gitoxide (gix) is a
command-line interface (CLI) to access Git repositories. It’s
written to optimize the user-experience, and perform as good or
better than the canonical implementation.
It provides an idiomatic, lean, fast and safe pure Rust implementation of Git, and it also provides an easy and safe to use API in the form of various small crates for implementing your own tools.
grokmirror-2.0 is a Git mirroring software announced by Konstantin Ryabitsev on the Git mailing list. It was developed by kernel.org for their own use.
GitQlient, pronounced as git+client (/gɪtˈklaɪənt/), is a multi-platform Git client written with Qt, originally forked from QGit.
Related project is GitQlientPlugin, the project to build an integrated QtCreator plugin of GitQlient.
Graphtage: A New Semantic Diffing Tool is a command line utility and underlying library for semantically comparing and merging tree-like structures such as JSON, JSON5, XML, HTML, YAML, and TOML files. You can even compare files that are in two different formats. When paired with the PolyFile tool you can semantically diff arbitrary file formats.
Sourcetrail is a free and open-source cross-platform interactive source explorer, that helps you get productive on unfamiliar source code. It currently supports C, C++, Java and Python.
git-in is
a small Python script to import a message or an entire thread from a mailing list
to an IMAP mailbox. Very useful when you are not subscribed to the list but want to answer
an email from the list, while being be able to quote parts of the message.
This edition of Git Rev News was curated by Christian Couder <christian.couder@gmail.com>, Jakub Narębski <jnareb@gmail.com>, Markus Jansen <mja@jansen-preisler.de> and Kaartic Sivaraam <kaartic.sivaraam@gmail.com> with help from Ethan Rahn, Philippe Blain and Junio Hamano.