[Authentication] Spec review (part 1)

[Michael Leupold]

Hi Stef,

2011/2/27 Stef Walter <stefw at collabora.co.uk>:
> On 02/26/2011 11:33 PM, Valentin Rusu wrote:
>> On 02/09/2011 09:14 PM, Stef Walter wrote:
>>> On 01/30/2011 08:38 PM, Michael Leupold wrote:
>>>>> So this leaves us with either:
>>>>>
>>>>> a) A "Type" property
>>>>> b) An optional "Type" attribute
>>>>> c) A "type" member in the Secret-struct
>>>>>
>>>>> I like a) for its simplicity, b) seems kinda awkward with the only
>>>>> benefit being that you could search/filter items by type, c) seems to
>>>>> be the cleanest solution. It provides consistency as you'd always
>>>>> get/set the actual value and its type in one go. I'd be in favour of
>>>>> c).
>>>>>
>>> How does the attached patch look?
>> Ok, I took your changes into account inside ksecretservice sources.
>
> That's great. I'll work on implementing them in gnome-keyring. I need to
> figure out where to put this extra field in the old gnome-keyring file
> format.
>
> At some point we'll switch over to a fancy-schmancy new file format, but
> we're not there yet.
>
> Michael, do you have any comments on the patch?

No, looks fine to me.

We (as the individual daemon and library developers) should make sure
we use similar data-types so each of our clients can display most or
all of them (eg. gnome-keyring should be able to display KDE
applications' values to the user). We'll keep you posted which types
we'll be using to support legacy KWallet secrets.

Thanks and regards,
Michael