Re: [PATCH v2 00/18] Add `branch-diff`, a `tbdiff` lookalike

Hi,

On Fri, May 4, 2018 at 9:21 AM, Elijah Newren [off-list ref] wrote:

On Fri, May 4, 2018 at 8:34 AM, Johannes Schindelin
[off-list ref] wrote:

quoted

The incredibly useful `git-tbdiff` tool to compare patch series (say, to see
what changed between two iterations sent to the Git mailing list) is slightly
less useful for this developer due to the fact that it requires the `hungarian`
and `numpy` Python packages which are for some reason really hard to build in
MSYS2. So hard that I even had to give up, because it was simply easier to
reimplement the whole shebang as a builtin command.

tbdiff is awesome; thanks for bringing it in as a builtin to git.

I've run through a few cases, comparing output of tbdiff and
branch-diff.  So far, what I've noted is that they produce largely the
same output except that:

- tbdiff seems to shorten shas to 7 characters, branch-diff is using
10, in git.git at least.  (Probably a good change)

Sorry, a quick self-correction here:

tbdiff, when using an actual shortened sha, uses 10 characters.  But
when a patch doesn't have a match, tbdiff seems to use seven dashes on
one side in lieu of a shortened sha, whereas branch-diff will use 10
characters whether it has an actual shortened sha or is just putting a
bunch of dashes there.  So, this is definitely a good change.

- tbdiff aligned output columns better when there were more than 9
patches (I'll comment more on patch 09/18)
- As noted elsewhere in the review of round 1, tbdiff uses difflib
while branch-diff uses xdiff.  I found some cases where that mattered,
and in all of them, I either felt like the difference was irrelevant
or that difflib was suboptimal, so this is definitely an improvement
for me.
- branch-diff produces it's output faster, and it is automatically
paged.  This is really cool.

Also, I don't have bash-completion for either tbdiff or branch-diff.
:-(  But I saw some discussion on the v1 patches about how this gets
handled...  :-)