3 ms·
Meh. In such a case, you'll find that the older code (skipping the bulk change commit) does not have the (mis)feature you're looking for, and you'll have to go
by AceJohnny2 5y ago
Meh.
In such a case, you'll find that the older code (skipping the bulk change commit) does not have the (mis)feature you're looking for, and you'll have to go the extra step of looking through the ignored commits.
It makes the case of finding changes in some commits a bit harder (but not hard), in exchange for reducing the noise from those same commits. Having been long hampered by ugly codebases that I didn't dare touch for fear of "blame" poisoning, I think this is a net positive overall.
- AceJohnny2 5y agoI actually encountered a similar situation years ago, where I tried to bisect for a regression. Problem was, old code wasn't testable for my regression, and I had to apply a patch to it to be able to automatically test it. I kept getting nonsensical results out of the bisect however. I later realized that my patch to make old code testable introduced the very regression I was looking for!