> b) so, you agree with the blog? "So, any realistic interesting attack vector therefore relies on a collision attack,"

My reading of the blog is that they are dismissive of collision attacks. In context of git, i disagree. I think there are plausible attack scenarios involving collisions, or at least, just as plausible as second pre-image.

If you mean do i agree with the blog that impossible attacks aren't possible? well yes obviously, but i think that goes without saying.