From: Junio C Hamano <hidden> Date: 2016-06-15 22:43:45
Thanks. This looks obviously correct and tempts me to apply
directly to 'master'.
There are similar logic on the gitattributes side (attr.c) which
might benefit from such an optimization as well. In the longer
run, we might even want to introduce 'ignored' attribute to the
gitattributes mechanism and make the unified result to supersede
gitignore.