Hi!

1. Those kind of changes would be welcome, and I don't have a strong preference between putting them in core code or an extension point.  Often core code is simpler, and can be controlled with a config value still.

2. Existing extension points are pretty stable, yes.

3. Its good you're considering project health.  Not only is the recent development from a small group of us, we are all employed at the same place.  That's not ideal for an OSS project, especially at the ASF, but its the reality.  Not many organizations run their own "forge" and new contributors are very infrequent.  Your potential contributions would be the first like it in many years, and still aren't really going to add new life into the project.  I can't say for sure what Allura's future is, but to be completely honest its likely to dwindle more.  Its probably not a good long-term choice.

-Dave

On 9/25/26 10:33 AM, Piotr P. Karwasz wrote:
From: Piotr P. Karwasz <[email protected]>
To: [email protected]
Subject: ASF Security Team evaluating Allura as a security-report tracker
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 8bit

Hi all,

We're looking for a replacement for how the ASF Security Team tracks
inbound vulnerability reports, and Allura is our leading candidate.
Before we commit, we'd like your view on two things: whether the changes
we'd need could be accepted upstream, and whether building on Allura is
sustainable from your side.

The problem

Today we track reports sent to [email protected] with Gmail labels,
plus a small in-house dashboard that gives each PMC access to its own
reports. This no longer scales. We want a proper tracker: one private
Allura instance, one neighbourhood, and one project for each of the ~200
PMCs. Each report becomes a ticket, and PMC members can see only their
own project's tickets.

What we found

The tracker side fits well: custom fields, configurable statuses,
labels, search, history and the REST API all do what we need. The
entry-point architecture also means we shouldn't need a fork. We'd still
need to build some things, and a few of them are generic enough that
they may belong upstream:

- Inbound mail routing. Allura routes incoming mail only by recipient
address, so mail sent to a single fixed address such as
[email protected] has nowhere to land. We'd write a gateway that
matches follow-ups to tickets by Message-ID. A small, generic hook in
the mail-handling path would make this much cleaner than re-injecting
mail into the SMTP listener.
- Group-to-role mapping. Allura's LDAP support handles authentication
only. We'd write an ASF OAuth provider that syncs PMC membership into
project roles at login. That provider is ASF-specific and would probably
stay in our own repository as a plugin. A generic "sync external groups
to roles" hook in the auth provider API might be useful upstream, though.
- Read-access audit logging. We need a record of who viewed a restricted
ticket, especially when the viewer isn't a member of the project.
- Tracker statistics. We need a chart of open/closed reports over time
for each project.

Our questions

1. Would you be open to accepting changes like these upstream, and do
you have a preference between core changes and new extension points?
2. Would you consider the extension points we'd rely on (Application,
auth providers, template overrides, the mail handler) stable enough to
build on?
3. How do you see the project's health over the next few years? Looking
at the history, the project is clearly active, and the recent hardening
work under #8624 was encouraging. Most commits come from a small group
of people, though. Since the Foundation's whole security intake would
depend on this system, we want to ask openly whether you think this is
sustainable, rather than guess.

To be upfront: we can't promise a steady stream of contributions. Our
main work is elsewhere. We would upstream the things we build if you
want them, report bugs, and report any security issues we find through
the usual process. Anything not accepted upstream we'd maintain
ourselves as plugins.

Any thoughts are welcome, including "this isn't a good fit."

Thanks,
Piotr

--
Dave Brondsema

Reply via email to