Provide guidelines for mitigation algorithms #241
Conversation
This patch is providing guidelines on numerical values to select for the mitigation algorithms parameters. [1] [1] w3c#197 (comment) Fixes: w3c#240
index.html
Outdated
| <p><i>This section is non-normative.</i></p> | ||
| <p> | ||
| Based on implementation experience, implementers are advised to apply the mitigation | ||
| to a randomized time value within a range between 120000 milliseconds (2minutes) and 240000 milliseconds (4 minutes). |
There was a problem hiding this comment.
| to a randomized time value within a range between 120000 milliseconds (2minutes) and 240000 milliseconds (4 minutes). | |
| to a randomized time value within a range between 120000 milliseconds (2 minutes) and 240000 milliseconds (4 minutes). |
There was a problem hiding this comment.
I ll make the change in my commit. easier
index.html
Outdated
| Based on implementation experience, implementers are advised to use: | ||
| <ul> | ||
| <li> | ||
| a range in between 300000 milliseconds (5 minutes) and 600000 milliseconds (10 minutes) for |observer|.{{PressureObserver/[[ObservationWindow]]}}. |
There was a problem hiding this comment.
Editorial nitpick: the |variable| notation is generally used in algorithms. observer has not been defined here. I'd just say something along the lines of "a range [...] for PressureObserver's [[ObservationWindow]] internal slot".
There was a problem hiding this comment.
Approved with @rakuco's editorial nitpicks addressed.
As a general advise, to make an analogy, informative sections for web specs are similar to comments in code.
What follows is that anything that's not testable should be an informative note.
There was a problem hiding this comment.
Reading #197 (comment), my impression is that Pete suggested that the spec should have normative ceiling and/or floor values for the intervals described here (this would also help when writing web tests).
Is it really not possible at this point to make any of the values normative?
|
We do have some wpt tests for [[MaxChangesThreshold]] and [[PenaltyDuration]], checking that the random [[maxChangesThreshold]] number choosen is in the range and that penalty is happening during the [[PenaltyDuration]] range. the [[observationWindow]] is not as important, it is used to give more randomness to the process, so it can stay non-normative. |
I think it's fine to define them normatively and have a note saying the values are based on implementation experience and are therefore subject to change. Changing the values in the future then becomes like any other normative change with user-visible effects, i.e. get buy-in from all implementers (not hard in this case yet since there's just one implementation) and adjust existing web tests. |
Yes, definitely! |
|
I think this approach sounds great, thank you for the work! |
|
Thank You @pes10k for helping us harden this specification further! |
This patch is providing guidelines on numerical values to select for the mitigation algorithms parameters. [1]
[1] #197 (comment)
Fixes: #240
Preview | Diff