mailing list archives
Re: running the distros lists
From: Josh Bressers <bressers () redhat com>
Date: Mon, 19 Mar 2012 11:12:11 -0400 (EDT)
----- Original Message -----
Josh, Kurt, all -
On Thu, Mar 15, 2012 at 08:33:41AM -0400, Josh Bressers wrote:
This task strikes me as a timesink that results in minimal value. It's
in the best interest of all list members to fix issues in a timely
manner. The historic vendor-sec never had a serious problem with CRD.
They were often outside of the current proposed 14 day span, but that's
just a reality of how things go. We all have more work to do than time,
so some things are bound to suffer.
This makes some sense, but:
1. I think that merely proposing a CRD for every reported issue right
away is not a timesink. I could probably be doing it myself, but I'd
like to get other list members more involved in running the list in
general, I feel that I am already contributing by administering the list
setup (so it's not my turn to contribute more), and I think that someone
with a vendor affected by a larger percentage of the issues is in a
better position to propose CRDs.
Proposing the date isn't the issue. If you want to continue with your 14
day embargo, that's trivial. The problem is chasing that. Who is
responsible for ensuring everything is progressing as expected?
2. It is bad that embargo periods on the historic vendor-sec were often
pretty long. I don't want the same to be happening on the distros list.
Yes, the reality is that some issues will end up being embargoed for
longer than many of us would have liked, but let's try to reduce the
embargo periods whenever we can and to the extent possible. Proposing a
CRD early on is a step in that direction.
The biggest problem vendor-sec had was low priority issues would just sit
and rot. Many probably never should have had an embargo to begin with. This
is where a relatively short default embargo date is useful. If nobody
cares, let it go public. At the same time, if something is very important,
they can ask for an extension.
The issue of course becomes how does something become public then if nobody
Trying to get a couple people to do this is going to be a losing
Maybe, but why (in your opinion)?
My above reasoning is why I say this.
We need to think about how to best let the list police itself.
Do you have any specific proposal?
This is where things get tricky using just a list. Ideally, things get a
default embargo date, then they open up at that time (sadly with a list,
this isn't automatic). If someone needs additional time, they request it,
again though, who will track this?
What I am thinking is that with only a few of us being responsible for
doing some well-specified things we will actually have to be doing those
things or be prepared to explain why we failed, apologize for that, and
adjust our actions going forward.
On the other hand, with the list membership at large being responsible
for running the list in accordance with certain goals and policies, no
one in particular is responsible. When things go wrong, there's not much
we can do - no one in particular feels like it's his/her fault, no one in
particular is going to change anything going forward. We can only choose
between ignoring the problem, calling it a non-problem, shutting the list
down, or switching to the "few responsible list members" scheme that I am
proposing now. I simply see no other options.
This is where things start to get tricky. I think about this problem a lot.
I think the fundamental issue is that we lack an open source security
community. We have a bunch of tribes, but no actual community. Until we
start to foster such a group, nothing is going to change. Embargoes are
really just a symptom of a much larger issue.
It certainly won't be 100% perfect, but I think even an 80% reasonable
CRD agreement goal is acceptable (perfect is the enemy of the good
comes to mind).
Maybe, but why can't we have a CRD proposed within a day in 100% of
cases? Not all of those initially proposed CRDs will be agreed to, but
at least proposing them without delay is desirable, I think.
The only way 100% is possible if there is automation. Humans can't do 100%.
I would suggest some reasonable guidelines (that are not enforced
strictly), then see how things go for a while. If it's deemed
unacceptable, a better solution can be worked on.
We already have what I think are reasonable guidelines. We see how
things are going: many of the reported issues linger without even a
proposed CRD for days. I find that unacceptable because at least in some
cases it leads to increased total embargo periods. Thus, I think we're
ready to discuss and hopefully arrive at a solution.
I think the solution is to start creating an open source community. Of
course this is far easier said than done. No doubt we all have some ideas.
I think a good first step would be getting a bunch of folks in the
#oss-security channel on Freenode. We can start talking and figure out
where to go from there. There are some tools available to us now that were
not a few years back (some social media type things). So far I think the
wiki and mailing lists are a start, but we could make all this so much
more. We have an immense amount of talent out there, if we start
collaborating, I expect great things to happen.