On 20.11.14 23:37, Junio C Hamano wrote: > > --- > At a glance it is very hard to see what we might be _losing_ with this > change that claims to "add" new kinds of tests on top of existing ones. > > I am guessing that add-check-warn roughly corresponds to the old > create-file-in-repo but they have different calling conventions, or > something? > > Perhaps split it into two patches (or more), each of which does one > thing and one thing well? I suspect that even with a two-patch > split (e.g. the first of which only renames the function without > adding the new "grep in error messages that could be localized and > give false failures" code, and the second adds the lf/crlf stuff) > might make this at least readable. > > I dunno. > We shouldn't loose anything. The diff is hard to read, as some code is re-defined and re-used (and a diff side-by-side looks nicer than the patch) I will come back with a new commit message, which should explain things better ( or a 2-stepped patch) The long term idea is to improve the gray areas in convert.c, and to do that we need a reliable test frame work, to see what is improved or broken. -- To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html