Re: [PATCH v2 0/5] Git filter protocol

On Fri, Jul 29, 2016 at 07:43:49PM +0200, Lars Schneider wrote:

Here is the reject case (non-streaming):

git> smudge
git> <filename>
git> <size>
git> ...pkt-lines...
git> pktline-flush

git< 0
git< reject


Do you see a problem with this approach?

Only that it seemed a little weird to me to have to write a meaningless
"0" when "reject" covers the situation entirely. I don't think it's
wrong, though (and even in some ways right, because it decouples the
meaning of "reject" from the syntax of parsing, but I think it's OK for
the protocol parser to understand the difference between success and
failure codes).

-Peff