![](https://secure.gravatar.com/avatar/8a0cdd54b32f895fdf86d7aafd41d1fd.jpg?s=120&d=mm&r=g)
- Barry Warsaw <barry@list.org>:
How do we do it? Do I get write access to Mailman wiki?
You should have write access just by virtue of having an account on the wiki. There are only a few pages that aren't generally writable by every logged in user. If you're having a problem with a specific page, let me know.
I'll give it a try later.
We've thought about different client technologies too. That's the client technology part I wrote about in the wiki.
Which we didn't discuss was fully authenticated access for the REST server by design. If I understand this correctly than any party that is able to communicate with the REST server will have full admin access to Mailman's data model. In other words: It's upon any REST client to protect the REST server from abuse.
That's basically correct.
I feel a little uneasy not having the server control that itself unless we find a good way to control who may connect to the server or the server is able to identify valid clients by some client identity (ACL).
It depends on whether we view the REST API as a user feature or an admin interface. I've always thought about it as the latter, but I'm open to
It's probably both, depending on the users role.
other opinions. OTOH, I think there's a lot of functionality that a privileged process could need, that the general public won't need at all.
That's what I think, too.
Another way to think about it is that there doesn't need to be just one REST API.
Yes and I think this would make maintaining code, setting the whole system up and configuring it more complicated.
Currently one REST server that uses a role model to determine access level to MM's data model seems the best approach to me. I am open to suggestions.
What this means though is that when you deploy Mailman's REST interface, you must take care to protect it. You wouldn't want to expose it to the internet for example. You'd want to make sure that its interface is
Exposing it to the internet is a typical use case in my eyes e.g. run the server on the internet, but control it from a different host. I can see mailman providers offering access to their MM server to customers who integrate their client on their servers - on the internet.
accessibly on via your data center, or via localhost if you were running a turnkey standalone system.
I was thinking of TLS client/server authentication for open networks. Not that I have spent time yet to find out if Python (REST) tools provide such functionality - I am sure it does, but given my low Python experience, I'd rather verify...
I'm not sure about this either.
We should check. Client/server communication will send/receive personal data that IMHO should always be protected during transport regardless of the REST data access control model we choose.
p@rick
-- state of mind Agentur für Kommunikation, Design und Softwareentwicklung
Franziskanerstraße 15 Telefon +49 89 45227227 81669 München Telefax +49 89 45227226
Amtsgericht München Partnerschaftsregister PR 563