On 15.12.20 09:11, Dominic Jäger wrote:
> On Mon, Dec 14, 2020 at 01:33:25PM +0100, Fabian Grünbichler wrote:
>> I think the argument is that the associated format already enforces that
>> the ID is >= 100, so that the API allows less has no practical effect
>> except confusing users that read the man page/help output/api dump.
>
> This is exactly what I meant. But I'm not 100% sure if some schema magic
> could
> happen that circumvents the pve-vmid format, so I can certainly start
>
FWIW; I never questioned that and this was 100% clear to me, but it still
does not help if we have some use here, even if only legacy or restore wise
or what not, touch such things can easily break things in a subtle way, and
only get noticed in months or even years, we had such stuff a few times already,
so I'd be really cautious here - adding the "actual enforced limit" into the
description can also help.
I mean, I really do not think this crucial limit slipped and was left as is
for years...
> On Mon, Dec 14, 2020 at 12:52:35PM +0100, Thomas Lamprecht wrote:
>> maybe you could take another look if you find
>> some case where values <100 are used for some special handling?
>
> looking for this.
>
>>>> diff --git a/src/PVE/JSONSchema.pm b/src/PVE/JSONSchema.pm
>>>> index 29ada5b..b4596d3 100644
>>>> --- a/src/PVE/JSONSchema.pm
>>>> +++ b/src/PVE/JSONSchema.pm
>>>> @@ -58,7 +58,7 @@ sub get_standard_option {
>>>> register_standard_option('pve-vmid', {
>>>> description => "The (unique) ID of the VM.",
>>>> type => 'integer', format => 'pve-vmid',
>>>> - minimum => 1
>>>> + minimum => 100,
>>>> });
>>>>
_______________________________________________
pve-devel mailing list
pve-devel@lists.proxmox.com
https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-devel
