On Fri, Apr 15, 2011 at 03:09:08PM -0400, Jeff King wrote: > On Fri, Apr 15, 2011 at 10:46:28AM -0700, Junio C Hamano wrote: > > > Shawn Pearce <spearce@xxxxxxxxxxx> writes: > > > > > On Thu, Apr 14, 2011 at 23:22, madmarcos <fru574@xxxxxxxxxxx> wrote: > > >> is there a git command that can capture in a separate file the packfile that > > >> is generated and transmitted from a "want"ed branch during the > > >> git-upload-pack command? I need a good sample to study. > > > > > > No. > > > > > > But you can create an input file yourself, the pkt-line format is > > > pretty simple. Pipe the input to git-upload-pack, and it will spit > > > back the response. :-) > > > > If you are running 1.7.5-rc0 or newer, GIT_TRACE_PACKET may also be of > > help. > > Sadly, no. I punted on outputting the actual pack since it is big and > binary, and I was only interested in the ref negotiation. My eventual > plan was that you could do: > > GIT_TRACE_PACKET=2 GIT_TRACE_PACKET_PACK=/tmp/foo.pack \ > git clone ... Actually, thinking on it more, this is kind of stupid. The pack that we fetch will end up in .git/objects/pack, so it's not that interesting to dump (though I suppose in a fetch setting, you might want to actually peek at the thin pack before it gets resolved). More interesting is for the sender to dump the pack they send, which would mean instrumenting pack-objects. Or, as Shawn suggested, poking upload-pack yourself using a conversation you read from GIT_TRACE_PACKET. Which is probably what you meant when you suggested it in the first place. So I'll shut up now. -Peff -- 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