Re: [GSoC] Introduction

[Johannes Schindelin <Johannes.Schindelin@gmx.de>]

Hi Rohit,

On Sun, 24 Feb 2019, Rohit Ashiwal wrote:

> Hello

Hi there!

> I am Rohit Ashiwal, a computer science and engineering student from
> IIT Roorkee. I am a git user for 2 years now and want to contribute
> towards its development.

Great!

> I have read the microproject list available [here](1) and project
> titled `Use test_path_is_*  functions in test scripts` caught my eye.
> I want to work on this one if anyone is not working on it (I was not
> able to find related threads on public-inbox).

Go for it ;-) The microprojects are basically a straight-forward way to
introduce prospective GSoC students to Git's code contribution process.

In that sense, while it would be better for students not to have
overlapping microprojects, it is not the end of the world, either.

In your case, it is relatively easy to figure out that no other student
mentioned this project on the Git mailing list:
https://public-inbox.org/git/?q=test_path_is_file

> I am somewhat familiar with git's code as I was previously working on
> git-for-windows and I have solved 1 [issue](2), which is a reason for
> my concern. Should this count as a microproject? As the webpage
> instructs us to work only on 1 microproject.

True. You could brush up that pair of patches to teach `git archive` to
avoid spawning `gzip` when generating `.tar.gz` archives, and submit them
using the process suggested in https://git-scm.com/docs/SubmittingPatches.

(You might want to make it easier on yourself and use GitGitGadget, which
takes out the challenge of sending patches in the required format via
email, see https://gitgitgadget.github.io/ for details.)

On the other hand, it might make more sense to start afresh with the
test_path_is_* microproject, as I already took custody of your patches and
planned on submitting them (via GitGitGadget) right after v2.21.0 comes
out.

> If you disregard that contribution, here is my approach to solving
> this microproject: search for `test -f` in `t/**` and replace all
> occurrences with `test_path_is_file` and similarly for other mentioned
> commands. Is it correct?

This is a good first step, indeed. You may want to be careful about it,
though, as there *might* be instances where replacing the `test -f` would
not make sense (such as instances where scripts are written via
`write_script`, and those scripts would not have access to the
`test_path_is_file` function).

You probably also want to inspect instances of `test -e` (which
technically only ask whether a path exists, whether it is a file, a
directory, or a symbolic link), as it might be possible that some tests
use that invocation when they really want to ask whether a file exists.

Then, you want to also look for `test -d` for the `test_path_is_dir`
conversion.

Good luck,
Johannes