Re: [RFC PATCH] btrfs-progs: Quiet mode for btrfs-receive

[David Sterba <dsterba@suse.cz>]

On Wed, Feb 20, 2019 at 09:16:04AM +0000, Steven Davies wrote:
> Provide an option in `btrfs receive` to suppress the informational
> messages when writing files.

Thank, looks ok to me, the 'receive' command already has the verbosity
option so it makes sense to have both.

> Signed-off-by: Steven Davies <btrfs@steev.me.uk>
> ---
>   cmds-receive.c | 67 ++++++++++++++++++++++++++++++--------------------
>   1 file changed, 40 insertions(+), 27 deletions(-)
> 
> diff --git a/cmds-receive.c b/cmds-receive.c
> index 3888149a..96411bac 100644
> --- a/cmds-receive.c
> +++ b/cmds-receive.c
> @@ -52,7 +52,11 @@
>   #include "send-dump.h"
>   #include "help.h"
> 
> -static int g_verbose = 0;
> +/*
> + * Default is 1 for historical reasons, changing may break scripts that 
> expect
> + * the 'At subvol' message.
> + */
> +static int g_verbose = 1;
> 
>   struct btrfs_receive
>   {
> @@ -194,12 +198,14 @@ static int process_subvol(const char *path, const 
> u8 *uuid, u64 ctransid,
>                  goto out;
>          }
> 
> -       fprintf(stderr, "At subvol %s\n", path);
> +       if (g_verbose) {
> +               fprintf(stdout, "At subvol %s\n", path);
> +       }

This changes stderr to stdout.

> 
>          memcpy(rctx->cur_subvol.received_uuid, uuid, BTRFS_UUID_SIZE);
>          rctx->cur_subvol.stransid = ctransid;
> 
> -       if (g_verbose) {
> +       if (g_verbose >= 2) {
>                  uuid_unparse((u8*)rctx->cur_subvol.received_uuid, 
> uuid_str);
>                  fprintf(stderr, "receiving subvol %s uuid=%s, 
> stransid=%llu\n",
>                                  path, uuid_str,
> @@ -263,12 +269,14 @@ static int process_snapshot(const char *path, 
> const u8 *uuid, u64 ctransid,
>                  goto out;
>          }
> 
> -       fprintf(stdout, "At snapshot %s\n", path);
> +       if (g_verbose) {
> +               fprintf(stderr, "At snapshot %s\n", path);

And this stdout -> stderr.

That's inconsistent in the original code too, that may need some care to
fix. The stderr is used for all the verbosity levels and it's been like
that since the beginning. I doubt that it has a good reason and that
everybody does 'btrfs receive 2>&1' anyway, but we can never be sure
with that.

So for sake of backward compatibility I'll fix it to keep using the same
output streams and add to devel. Thanks.