monotone-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Monotone-devel] Re: read/write permissions


From: Ben Walton
Subject: Re: [Monotone-devel] Re: read/write permissions
Date: Fri, 26 Jan 2007 20:56:48 -0500

I think the project_t work that is being done currently might make
this a little more sane.  Personally, I've shied away from mixing
multiple projects in one db in favour of a ~/monotone directory
containing one .mtn db per project.  It just felt cleaner to me.
YMMV.

-Ben

On 1/26/07, Boris <address@hidden> wrote:
On Fri, 26 Jan 2007 22:16:37 +0200, Graydon Hoare <address@hidden>
wrote:

> Boris wrote:
>> What happens if a user who has only read-permission for branch A writes
>> something to branch B? As far as I see this is possible as
>> write-permissions are per database and not per branch. If branch B is
>> only for privileged users is it still protected? I'd say so as the new
>> user can't link his certificate into existing revisions in branch B. Is
>> this correct? But does everyone with read-permission to branch B see
>> now two heads?
>
> I believe at the moment that any database set up to reject the user
> writing to branch B will reject certificates from that user that bind a
> revision to B, and possibly reject the revision as well.
>
> Still, it's possible for the revision and/or cert to be transmitted by
> some other means. Using read or write permissions to control trust is
> the wrong thing to do; permissions are really just coarse measures to
> control visibility and abuse. The point of modeling commits with certs
> is that users at the ends of the system can decide what to trust
> themselves, independent of the trustworthiness of intermediaries in the
> transmission path.
>
> (We're hoping to put together a simpler and more useful bootstrapping
> trust system eventually. For the time being, users must define their own
> trust rules with lua hooks)

The reason I ask is that branch B contains source code and branch A tests.
A tester should be able to read/write branch A but not branch B. I can
restrict read-access but can't avoid that a tester writes to branch B. As
far as I understand I need to change get_revision_cert_trust() to make
sure that no code from a tester ends up in my workspace and in the release
version of the product.

Your argument that certs can be transmitted differently makes sense. But
then I wonder if my approach is wrong: Using different branches in the
same database for different projects doesn't seem to make much sense if
you want to restrict read/write access per project. I would need to put
one project in one database to do this (which would complicate matters
however if you have a central server you'd like everyone in the team to
synchronize with - no matter what project they work on).

Any ideas how to organize this?

Boris



_______________________________________________
Monotone-devel mailing list
address@hidden
http://lists.nongnu.org/mailman/listinfo/monotone-devel



--
---------------------------------------------------------------------------------------------------------------------------
Ben Walton <address@hidden>

Unanswered questions are far less dangerous than unquestioned answers.
- The Roadside Pulpit
---------------------------------------------------------------------------------------------------------------------------




reply via email to

[Prev in Thread] Current Thread [Next in Thread]