On Mon, Sep 12, 2016 at 5:11 PM, Junio C Hamano <gitster@xxxxxxxxx> wrote: > Stefan Beller <stefanbeller@xxxxxxxxx> writes: > >> From: Stefan Beller <sbeller@xxxxxxxxxx> >> >> --- > > "X can do Y" can be taken as a statement of fact (to which "so > what?" is an appropriate response), a desire (to which "then please > say 'make X do Y' instead" is an appropriate response), or a report > of a bug (to which "please explain why X should be forbidden from > doing Y" is an appropriate response). > > This is way under-explained. I think this is "make X do Y" kind, > and if so, please say so and possibly why it is a good idea to teach > X how to do Y. > > Thanks. Ok, I see the general pattern of your answers: Add more explanations. Answering for patch 01/10 as well as this one here: I want to propose an option to detect&color moved lines in a patch, which cannot be done in a one-pass over the diff. Instead we need to go over the whole diff twice, because we cannot detect the first line of a line pair that got moved in the first pass. So I aim for * collecting all output into a buffer as a first pass, * as the second pass output the buffer. So in a later patch I will split up the emit_line_* machinery to either emitting to options->file or buffering if we do the 2 pass thing. To make sure the 2 passes work correctly, we need to make sure all output is routed through the emit_line functions, and there will be no direct writes. Now that we will be using the emit_lines functions for non colored output as well, we want to pass in "no color" which I think is best done via NULL and then not calling the output of the color writes.