Re: [libvirt] [RFC] virCommandRun return error number
On 04/11/2013 07:48 AM, harryxiyou wrote:
On Thu, Apr 11, 2013 at 9:38 PM, Michal Privoznik
<mprivozn(a)redhat.com> wrote:
[...]
> Maybe you've pasted wrong code snippet, but the one you've pasted is
> correct. If virCommandRun() fails and returns -1; we return that value
> to the caller. And if the command doesn't fail, we call
> virStorageBackendSheepdogParseNodeInfo to parse output of the command.
>
Sorry, i have asked for a stupid bug. Let us check following one.
[...]
114 static int
115 virStorageBackendSheepdogRefreshPool(virConnectPtr conn ATTRIBUTE_UNUSED,
116 virStoragePoolObjPtr pool)
117 {
118 int ret;
119 char *output = NULL;
120 virCommandPtr cmd;
121
122 cmd = virCommandNewArgList(COLLIE, "node", "info",
"-r", NULL);
123 virStorageBackendSheepdogAddHostArg(cmd, pool);
[...]
virCommandNewArgList may return ‘NULL’, so i think we should check
this condition, which we need not do following stuffs, right?
Not necessary. virCommand is DESIGNED for streamlined usage, so that it
is much easier to read how the command is constructed without being
distracted by error checking in the caller every step of the way. As a
virCommandPtr has no semantic impact until it is run, it is sufficient
to delay error checking until the caller is actually ready to run the
command. Therefore, we wrote virCommandRun() to specifically check for
NULL, and report an error at that time, so that the caller need not
worry about virCommandNew* returning NULL.
No bug here.
--
Eric Blake eblake redhat com +1-919-301-3266
Libvirt virtualization library http://libvirt.org
Attachments:
- signature.asc (application/pgp-signature — 621 bytes)