git-rev-list --merge-order bug?
I've noticed that git-rev-list will sometimes return a different *number* of commit objects depending on whether the "--merge-order" switch was used.
For example:
Get a copy of Linus' git git repository from rsync.kernel.org and try the following:
git-rev-list ee28152d03f2cf4b5e3ebc25f7f03f9654d3aa0d \ ^aa03413467a2f2ada900817dc2a8e3904549b5fe |wc -l
git-rev-list ee28152d03f2cf4b5e3ebc25f7f03f9654d3aa0d \ ^aa03413467a2f2ada900817dc2a8e3904549b5fe --merge-order |wc -l
The first git-rev-list returns 4 commits and the second returns 304.
Those commits are dated 2005-05-03 and 2005-05-02. If you use gitk, search for ``builtin'' which will jump to the merge base of the 2 branches.
Quoted 9 lines
git-rev-list ee28152d03f2cf4b5e3ebc25f7f03f9654d3aa0d \ ^aa03413467a2f2ada900817dc2a8e3904549b5fe |wc -lgit-rev-list ee28152d03f2cf4b5e3ebc25f7f03f9654d3aa0d \ ^aa03413467a2f2ada900817dc2a8e3904549b5fe --merge-order |wc -lThe first git-rev-list returns 4 commits and the second returns 304.
I'll look at it.
It appears to do in cases that look like this:
A
| B | | | /
C
git-rev-list B ^A --merge-order
jon.
Signature · 2 lines
homepage: http://www.zeta.org.au/~jon/ blog: http://orwelliantremors.blogspot.com/