On Mon, Dec 07, 2020 at 03:47:39PM +0800, Shi Lei wrote:
On 2020-12-05 at 02:17, DanielP. Berrangé wrote:
>On Fri, Sep 04, 2020 at 11:35:29AM +0800, Shi Lei wrote:
>> Signed-off-by: Shi Lei <shi_lei(a)massclouds.com>
>> ---
>> src/conf/domain_conf.c | 272 ++---------------------------------------
>> src/conf/domain_conf.h | 37 +++---
>> 2 files changed, 26 insertions(+), 283 deletions(-)
>>
>> diff --git a/src/conf/domain_conf.c b/src/conf/domain_conf.c
>> index b3ec111..20d731b 100644
>> --- a/src/conf/domain_conf.c
>> +++ b/src/conf/domain_conf.c
>> @@ -877,6 +877,7 @@ VIR_ENUM_IMPL(virDomainGraphicsVNCSharePolicy,
>>
>> VIR_ENUM_IMPL(virDomainGraphicsSpiceChannelName,
>> VIR_DOMAIN_GRAPHICS_SPICE_CHANNEL_LAST,
>> + "none",
>> "main",
>> "display",
>> "inputs",
>> @@ -14431,13 +14432,14 @@ virDomainGraphicsRDPDefParseXMLHook(xmlNodePtr node
>> diff --git a/src/conf/domain_conf.h b/src/conf/domain_conf.h
>> index df84763..f27f429 100644
>> --- a/src/conf/domain_conf.h
>> +++ b/src/conf/domain_conf.h
>> @@ -1584,6 +1584,7 @@ struct _virDomainGraphicsAuthDef { /* genparse,
genformat:separate */
>> };
>>
>> typedef enum {
>> + VIR_DOMAIN_GRAPHICS_SPICE_CHANNEL_NONE = 0,
>
>I'm not sure why this extra enum field needs to be added ?
>
>IMHO we don't really want to have such extra values except in
>a few special cases where we need to track some "default"
>explicitly.
There're two forms of enum in the code.
(1) Start with a 'default' or 'none' item, e.g., virDomainDiskIo
The virDomainDiskIo has VIR_DOMAIN_DISK_IO_DEFAULT which is ZERO-indexed.
When parsing it, the code is like:
if ((tmp = virXMLPropString(cur, "io")) &&
(def->iomode = virDomainDiskIoTypeFromString(tmp)) <= 0) {
virReportError(...);
return -1;
}
It checks XXXTypeFromString() ** <= ** 0, because the form <...
io='default'>
is illegal.
(2) Without a 'default' or 'none' value, e.g., virDomainDiskDevice
When parsing it, the code is like:
if ((tmp = virXMLPropString(node, "device")) &&
(def->device = virDomainDiskDeviceTypeFromString(tmp)) < 0) {
virReportError(...);
return -1;
}
It just checks XXXTypeFromString() ** < ** 0, because the enum's ZERO-Value
item
is valid.
To handle these two situations, I add extra "default" for case #2 and unify the
form of enums,
so that it's easier to implement the generator.
Another solution is to add an extra directive to indicate whether a enum has a
'default' item,
so that the generator can decide between '<=' and '<'. Perhaps it
is more safe and reliable.
Looks like there are about 200 enums, 43 use _NONE and 56 use _DEFAULT,
and the remaining 100 use neither.
Maybe we can do without the need to add a directive, if we change all cases of
NONE to DEFAULT, and just make the generator look at whether _DEFAULT is present
or not ?
Regards,
Daniel
--
|:
https://berrange.com -o-
https://www.flickr.com/photos/dberrange :|
|:
https://libvirt.org -o-
https://fstop138.berrange.com :|
|:
https://entangle-photo.org -o-
https://www.instagram.com/dberrange :|