layout | title | navbar |
---|---|---|
default |
Outreachy Round 15 (September 2017) |
false |
Git is participating in Round 15 of Outreachy, an internship program that gives people from under-represented groups the opportunity to work on Free and Open Source Software projects. You can find out more about the program, including eligibility, at its homepage.
This is the landing page for Git's participation in Round 15 of Outreachy, which will run from December 2017 to March 2018. Applications must be received by October 23rd; the full schedule and application instructions are available at the Outreachy site.
If you're considering applying for the internship, please look over the project list below, and get in touch with the community! Most communication happens via the mailing list, but you may also find people on IRC. Details for both can be found on our community page. You can email mentors directly if you want to discuss a potential project, but we encourage cc-ing the mailing list so that the community can benefit from the discussion.
- Jeff King <peff@peff.net>
- Christian Couder <christian.couder@gmail.com>
Note that this is just a list of suggested projects; we are happy to hear proposals for new projects. If you would like to propose an idea, please make sure to solicit feedback from the mailing list and mentors.
- Language: C
- Difficulty: medium to hard
- Possible mentors: Jeff King
Git frequently has to walk the graph of commits in order to show git log
output, compute merge bases, find which tags contain which commits,
and so on. It does so by inflating and parsing each commit object from
disk.
By caching the graph data in an efficient, pre-parsed form, we can speed up these traversals. Moreover, we can pre-compute properties of each commit (such as its generation number in the graph) to allow further optimizations.
- Language: C
- Difficulty: medium
- Possible mentors: Jeff King
Many Git commands take a --format
option to allow you to specify a
custom output format, but there are at least three distinct formats:
git log
output of commitsgit for-each-ref
output of refs (and commits they point to)git cat-file --batch-check
output of arbitrary objects
It would be less confusing if these all accepted the same syntax, and where possible provided access to the same items.
Most people don't start to work on open source with an ambitious three-month project. They work on small changes and get involved in the community by triaging incoming bug reports and requests, fixing bugs, and reviewing other people's code.
Rather than a large code project, the mentor would work with the intern to:
- learn to interact with the community and identify smaller projects worth working on
- complete a series of small projects, both from the list below and from bugs and ideas that come up on the mailing list
Some of the current "small bits" that could be worked on are below. Note that because these are smaller, some of them may be completed by others by the time the internship period begins. They're meant only as a representative sample:
- handling %(trailer:options) in for-each-ref
git-config
interface improvements (--default
and--stdin
options)- measure impacts of
pack_release_memory
(and possibly do away with it)
You may find more by searching for the "leftoverbits" tag in the mailing list archive.
- Language: C, shell (bash)
- Difficulty: medium
- Possible mentors: Christian Couder
When your project is strictly "new features are merged into trunk, never the other way around", it is handy to be able to first find a merge on the trunk that merged a topic to point fingers at when a bug appears, instead of having to drill down to the individual commit on the faulty side branch.
See:
- https://public-inbox.org/git/20150304053333.GA9584@peff.net/
- https://public-inbox.org/git/4D3CDDF9.6080405@intel.com/
Searching the mailing list for "bisect --first-parent" might be helpful too.
In some cases, git bisect
may test too many merge bases, thus
slowing down the bisection (making it closer to linear than
logarithmic).
See: