I think you are doing something wrong.

You have not pulled or merged or ... the master branch. Most of these commits here we already accounted for.

That is not in and by itself a big problem, it's just that we could do this more elegantly.

It is worth trying one or two things with git. You can rename your current RcppArmadillo working directory and copy of the repo -- say,RcppArmadillo-previous (or add the date, or ...).

The create a fresh checkout. That corresponds to our master branch here.

Then try to get both aligned. You should be able to the the 'working' copy (ie 'previous' above) to just about the same set of commits -- differing only by yournew commits made since I merged the branch.

Next time around, rungit pull --rebase <remote-name> <branch-name> before starting on a PR. This should bring your branch up to the present base without a "Merge branch 'master' of github.com:user/repo"

SeeAtlassian's Git Rebasing for more.

Why rebase and not just merge? The language about 'new commits' tends to confuse me because I do want the existing commits, not the old changes recommitted as new ones.

I think what I do in such a case is to

• pull the remote master (ie this repository) to be current
• branch cleanly ie with a freshgit checkout -b feature/newstuff which then contains all of master

But this is git so there are probably half a dozen ways...

That looks much better 💯

Looks really good. Two quick questions:

1. You mentioned

And there is a piece of codes using the algorithm from scipy.

Where is that? Does it need a attribution in the file header (i.e. something like 'Portions taken from file abc.py written by D, E and F and released under copyright G' ?)

2. So what did you do for the cleaner PR? Just a pull from our master, and push? It did the right thing as all the commits we already accounted

You can look at the referencehere. If attribution can be added, that would be great. It can be "Portions taken from filecsr.h written by SciPy developers and released under Copyright 2017 SciPy developers. "(?) I only find that on their website.

• git clone binxiangni/RcppArmadillo to local and git reset to the commit without all my previous commits.

• git pull from RcppCore/RcppArmadillo

• paste the three modified files to the local repo, git add and git commit

• git push -f to binxiangni/RcppArmadillo

1. My bad. I missed the reference. Yes. something like that would do. I may be hard to get concise author reference for SciPy.

2. I see. I didn't account for thegit push -f possibility.

All good.

All merged.

@binxiangni can you maybe write a short note at each of e#17 and#114 and close them, if appropriate, now that this is merged?