On Thu, Feb 22, 2024 at 04:17:35PM -0600, Eric Blake wrote:
Rather than changing captive nbdkit to export variables by default
(where we don't know if it will break clients of --run to suddenly
have more things in their environment variables), it is less invasive
to just call attention in our documentation that a run script that
relies on the environment must manually export variables.
Signed-off-by: Eric Blake <eblake(a)redhat.com>
---
docs/nbdkit-captive.pod | 7 ++++++-
docs/nbdkit-protocol.pod | 4 ++++
docs/nbdkit.pod | 2 ++
server/options.h | 1 +
tests/test-synopsis.sh | 1 +
5 files changed, 14 insertions(+), 1 deletion(-)
Shoot; I amended the patch incorrectly. This was intended to be just
a patch to nbdkit-captive.pod, and then I have a followup email for
the remaining changes to add a --no-structured-replies alias to
--no-sr, once it is fixed up to pass the testsuite.
diff --git a/docs/nbdkit-captive.pod b/docs/nbdkit-captive.pod
index 4f9740bc..94a9acb2 100644
--- a/docs/nbdkit-captive.pod
+++ b/docs/nbdkit-captive.pod
@@ -48,7 +48,12 @@ Running nbdkit captive under nbdsh for unit testing:
nbdkit memory 1 --run 'nbdsh -u "$uri" -c "print(h.pread(1,
0))"'
-The following shell variables are available in the I<--run> argument:
+Running an fio benchmark:
+
+ nbdkit -f null 1G --run 'export uri; fio fio.git/examples/nbd.fio'
+
+The following shell variables are available in the I<--run> argument
+(local to the run script unless you use C<export>):
=over 4
--
Eric Blake, Principal Software Engineer
Red Hat, Inc.
Virtualization:
qemu.org |
libguestfs.org