Re: [PATCH 5/5] Use start_comand() in builtin-fetch-pack.c instead of explicit fork/exec.

From: Junio C Hamano <gitster@pobox.com>
Date: Oct 4, 2007, 08:55 UTC
Message-ID: <7v8x6jb6of.fsf@gitster.siamese.dyndns.org>
In-Reply-To: <1191442180-15905-6-git-send-email-johannes.sixt@telecom.at>

Johannes Sixt <johannes.sixt@telecom.at> writes:

Show 22 quoted lines

> Signed-off-by: Johannes Sixt <johannes.sixt@telecom.at>
> ---
>  fetch-pack.c |   35 ++++++++++-------------------------
>  1 files changed, 10 insertions(+), 25 deletions(-)
>
> diff --git a/fetch-pack.c b/fetch-pack.c
> index d06b5ec..80268e1 100644
> --- a/fetch-pack.c
> +++ b/fetch-pack.c
> @@ -502,9 +503,12 @@ static int get_pack(int xd[2])
>  	char hdr_arg[256];
>  	const char **av;
>  	int do_keep = keep_pack;
> +	struct child_process cmd;
>  
>  	side_pid = setup_sideband(fd, xd);
>  
> +	memset(&cmd, 0, sizeof(cmd));
> +	cmd.argv = argv;
>  	av = argv;
>  	*hdr_arg = 0;
>  	if (unpack_limit) {

Your patch to this function makes status and pid unused variables, which I've fixed up locally.

The tip of your topic is currently queued to the tip of 'pu'; there were quite severe merge conflicts (textual conflicts in builtin-fetch-pack.c, and adjustment to transport.c for semantics change was also needed), so I ended up doing an evil merge there, which I am not very happy about.

I suspect the evil-merge's changes to builtin-fetch-pack to handle the connection to the index-pack process may be quite busted, but I ran out of time. Please check if the result makes sense, Ok?

I think Daniel and Shawn's git-fetch-in-C should graduate 'master' before this series. If you can re-send the series rebased on 2b5a06edca8f7237aad6464b349b79772024d2a2 (Restore default verbosity for http fetches.), it would be much appreciated.