On 05/05/2011 07:10 AM, Jiri Denemark wrote:
> They were not used anywhere so far so nobody noticed they are broken.
> ---
> src/util/json.c | 11 ++++++-----
> src/util/json.h | 4 ++--
> 2 files changed, 8 insertions(+), 7 deletions(-)
>
> diff --git a/src/util/json.c b/src/util/json.c
> index 0daeae9..df4771d 100644
> --- a/src/util/json.c
> +++ b/src/util/json.c
> @@ -491,12 +491,13 @@ int virJSONValueGetNumberDouble(virJSONValuePtr number, double *value)
> }
>
>
> -int virJSONValueGetBoolean(virJSONValuePtr val)
> +int virJSONValueGetBoolean(virJSONValuePtr val, bool *value)
> {
> - if (val->type != VIR_JSON_TYPE_NUMBER)
> + if (val->type != VIR_JSON_TYPE_BOOLEAN)
This bug was indeed embarrassing.
> return -1;
>
> - return val->data.boolean;
> + *value = val->data.boolean;
> + return 0;
> }
Technically, returning val->data.boolean was okay (0 or 1 is distinct
from -1), but then the user has the burden of checking for -1 before
assigning remaining values back to a bool.
> int virJSONValueGetNumberLong(virJSONValuePtr object, long long *value);
> int virJSONValueGetNumberUlong(virJSONValuePtr object, unsigned long long *value);
> int virJSONValueGetNumberDouble(virJSONValuePtr object, double *value);
And given the precedence of other functions that reserve the return
value for 0/-1 and pass the result by value through the argument pointer,
> -int virJSONValueGetBoolean(virJSONValuePtr object);
> +int virJSONValueGetBoolean(virJSONValuePtr object, bool *value);
I totally agree with this change.
ACK.
--
Eric Blake eblake@xxxxxxxxxx +1-801-349-2682
Libvirt virtualization library http://libvirt.org
Attachment:
signature.asc
Description: OpenPGP digital signature
-- libvir-list mailing list libvir-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/libvir-list