See also: IRC log
Florian: Anything else to add to the agenda?
kmag: I'd like to talk about standardizing IDs.
mikepie: Let's talk about extended attributes as well.
mikepie: We have some direct contacts what should help get the ball rolling by lightly prodding the right people.
https://github.com/browserext/native-messaging/wiki/Explainer:-the-need-for-native-messaging
Florian: This is a first incomplete draft of the join statement we mentioned during TPAC.
kmag: I am not sure this is really necessary.
Florian: Not sure either, but I think it's not wrong to ping people who worry about the whole web platform if we're doing something that might matter to more than to just ourselves.
mikepie: Looks like a good start,
will ping people at MS.
... What do we do once we have this?
Florian: I'll ping the TAG and see what they think of it.
Florian: Anything to speak about without aswan on the call?
mikepie: Not really.
Florian: Have we though about using URLs (as in including a domain name in there) for IDs? You'd still have to map the ID to the application, but at least we'd get a clean space without too much colision problems.
kmag: We'd need a way to verify
you own the domain, and some way to associate the ID with the
application... Seems complicated.
... Maybe we could instead share our ID database from the
respective vendor stores and allow registration on each store, but stores would
check with the other stores that the ID is free.
Florian: Doesn't that lock us out of native messaging when not using app stores for distribution? Like private extensions, or enterprise distribution...
kmag: I think it's ok. Enterprise works with its own rules anyway, and the single developer use case doesn't really involve distribution, so it is ok.
mikepie: How about having 2 ids?
a shared one that is unique across stores and one that is
unique in your own store?
... I guess that doesn't help.
Florian: Has our various companies already solved a similar problem, and can we talk to the people who did that?
mikepie: There's globaly unique Ids,
which deals with avoiding collisions, but it doesn't help with
central registration
... How about : when I submit to the FF store, and it generates
a unique ID for me. Then I use that ID to register in the MS
store.
Florian: But that allows for spoofing.
kmag: Only if the end user can do it, so we shouldn't allow.
mikepie: Can we combine that with the domain name id?
kmag: I don't see how.
Florian: Should we ping crypto / security / ssl / ssh / etc people? They should be more familiar than us with this kind of issues.
mikepie: Let's do that, I know some folks.
https://github.com/browserext/browserext.github.io/issues/12
<mikepie> https://mikepie1.github.io/browserext-1/LogoIdeas.png
mikepie: Number 2 is unfinished,
it will read ext instead of cxt.
... 1 and 6 may not work at small sizes
Florian I like 2 4 and 8
mikepie: 4 looks like XML.
kmag: I like 4 5 and 8.
mikepie: I'll clean up 2 4 5 and 8.
kmag: I'll also try to involve our designers.
<mikepie> https://mikepie1.github.io/browserext-1/
Florian: Any particular issue you want feedback / discussion on?
mikepie: Not sure, let's just
start with giving you an update:
... I merged the webdriver part.
... I also changed the naming convension, so that everything
starts with BrowserExt, and unified Info/etc into Details
... I added a script to facilitate navigating through
issues.
... so the thing I want input on is Issue 35, about some methods needing to be a POST
instead of a GET.
kmag: There are two methods that need to be POST, because they have side effects: “Get Browser Extension Context Menu Items” and “Select Browser Extension Context Menu Item”.
mikepie: OK. I'll be doing a new iteration on this tomorrow, and then the section should be ready for review.
Florian: Skipping the issues that are TODOs to yourself, it would be nice to register the issues that need discussions into github.
mikepie: Will do.
kmag: Should we provide regex for
comment removal?
... I'll register than into github, OK?
mikepie: Sounds good.
Florian: A while back we talked
about trying to reach out to more people and try to get them
involved, to comment on the spec.
... As we're starting to have a spec with issues to discuss,
are we at that point, or would it still be a waste of time for
most people to come at this point?
mikepie: I'd like to look at some
usage data to check first that we have the key APIs that people
will want to look at to avoid disapointing them, but other than
that yes, we should be able to now.
... I can look for extension vendors that can already use the
stuff we have
... If Andrey had some specific vendors in mind, that let's us
look at it from the other direction and ask them what they
think is missing.
mikepie: Let's try in two weeks.
Florian: Should we try and adjust the time for aswan?
mikepie: Yes, and for Shwetank as well.
Florian: Let's book the same time for now, and in parallel try and talk to them. We can adjust if they propose something that works better.
