Re: which algorithm do we use for copying buffer sources in dictionaries?

 > On Tue, Oct 23, 2018 at 2:03 AM =JeffH <Jeff.Hodges@kingsmountain.com
 >  <mailto:Jeff.Hodges@kingsmountain.com>> wrote:
 >
 > given
 > 
<https://github.com/w3c/webappsec-credential-management/issues/128#issuecomment-427607200>,
 >
 >  do we use the shorter..
 >
 > `<dfn>create a copy of any buffer sources' data</dfn>`
 >
 > ..one, or the..
 >
 > `<dfn>create a deep copy including any buffer sources</dfn>`
 >
 > ..one?
 >
 > Plus, what is the rationale for preferring one over the other?
 >
 > see also:
 > 
<https://github.com/w3c/webappsec-credential-management/issues/128#issuecomment-432133586>


On 10/23/2018 1:29 AM, J.C. Jones wrote:
 >
 > For what it's worth, I am in favor of `|<dfn>create a copy of any
 > buffer sources' data</dfn>` because it is closer to what actually
 > happens today in Firefox (at least).

Ok.

 > Particularly, everything in the dictionary is passed by reference
 > when we go async, except for buffers which are cloned.

hm, sorry, I'm confused (probably due to my ignorance), in part because 
<https://heycam.github.io/webidl/#idl-dictionaries> says "Dictionaries 
are always passed by value."

wrt "except for buffers which are cloned", are you referring to how the 
[=create a copy of any buffer sources' data=] works or something else 
(eg how FF works internally?)


 > A "deep copy" would suggest we should make some code changes to how
 > WebAuthn is called by Credential Management. At least how I read
 > it.

by "code changes" you mean actual FF code changes?

thx,

=JeffH

Received on Tuesday, 23 October 2018 17:24:53 UTC