Did he say it was for 'all purposes'? He said 50 characters for the first line is ideal for shortlogs, and 72 characters for more verbose logs, etc, when viewed in a terminal. He also said that encoding word wrap at write-time is better because only the author knows how to appropriately wrap. As he said, sometimes lines go longer, if the commit message includes program output that shouldn't be altered. So, you can't relegate word wrapping to the presentation layer without an unreasonably smart algorithm.
I interpret his statements to mean that the git/kernel projects' commit message standards are objectively better on all the ways we know how to measure (or at least, for all of the most common ways of looking at commit messages). So again, if you want to disagree, your choices are to point out a criterion that is not being considered, or how this system is not better at a known criterion.
I interpret his statements to mean that the git/kernel projects' commit message standards are objectively better on all the ways we know how to measure (or at least, for all of the most common ways of looking at commit messages). So again, if you want to disagree, your choices are to point out a criterion that is not being considered, or how this system is not better at a known criterion.