What everyone wants

Project Discussion
,
1

Many years ago I was sitting in the office of Denise Dumas, then the VP of all things Linux at Red Hat (“RHEL Engineering” at the time), lamenting that so many people want to know what’s on Red Hat’s mind, why can’t we just tell people what we want? “Why can’t we?” she asked, and thus was that “What Red Hat Wants” talk born. The first time was Flock 2015, in Cape Cod. The talk was good (better than the venue!) and has been repeated every year since, yet it has never quite achieved its purpose in my opinion.

In the clarity of hindsight, this is because the real underlying need was mutual understanding and interest. A single talk won’t do that. Indeed, even the best presentation is substantially unidirectional. Plus, when speaking for and to so many, the need for brevity battles the need for depth. If a single presentation won’t do the job, even multiple presentations won’t do the job, what would?

We need something more mutual, more engaging. And more open, perhaps with those internal RHEL folks you hear about, but rarely hear from. I can tell a community member why and how RHEL uses Gitlab + Jira, why CentOS Stream solves a real problem, and with equal enthusiasm, tell a RHEL insider why Fedora + Forgejo makes sense- and I’m just one person. The reality Red Hatters can share with authority and passion differs from person to person, what anybody is interested in differs from person to person, too. The magic is in sharing, finding common interests, and pursuing them together. In that spirit, here is an overview of what’s on my mind for “new” work in 2025, roughly aligned with my authority as the overall manager of CLE:

  1. Improvements to help people do the kinds of things they’re doing now, but it’s crufty to do or hard to maintain. That’s things like adopting improved software infrastructure such as Forgejo, that’s hardware infrastructure improvements leading to faster builds & tests, and working as community members to take advantage of these improvements with better processes and workflows
  2. Making new things possible, the likes of which we may not all yet even imagine. That’s finding and supporting a new FPL, expanding community opportunities to take part in more, and doing better at what people want but we haven’t figured out how to sustain.

These all require more sharing, more discussion, more opportunity to collaborate. I’ll be doing some of this personally, some of this with CLE, and sometimes encouraging more members of Red Hat to show up early and often. In this pursuit, I’d like to create some sort of 3-way interlock between leads in the RHEL, Fedora, and CentOS spaces. Meeting up at conferences is great, yet limiting. Having intentional, ongoing communication about works in progress, things not decided, that which may affect one another, or there would be a mutual interest is needed. When I think about it, it’s astonishing we don’t have something so fundamental.

18 Likes
2

These two categories of “new” work make sense to me. Although I also feel like commending the many folks in the Red Hat CLE / Fedora Infra team who already spend time working with community members, both as collaborators and mentors. Maybe instead of “new” work, it is enhancing and streamlining the efforts and work put in over many years. :muscle:

This seems like the main suggestion you are proposing. I think it makes sense, and this is a conversation that should happen openly and transparently. However, I am wondering more about what kind of conversations we should be having, or even what kind of conversations the community would like to see. I think there is sometimes a perception that “what Red Hat wants” and “what Fedora wants” are very much the same, but it is actually not always true.

So, if we want to have these conversations side by side, where should we begin? Is it focusing on infrastructure and tooling changes, like the upcoming migration to Forgejo? Is it about project strategy and engineering goals?

I think this makes sense and I want to be ally for making Red Hat and Fedora collaboration more transparent and positive.

3 Likes
3

I really do mean new work, as in major new undertakings. I don’t think this diminishes the work anybody does or has done, but maybe I’m I expressed the idea too generically to be understood and a few specifics would help: The most obvious new work is integrating a new git forge into the web of Fedora services. That’s not just CLE, there will be many community members involved, but it’s undeniably new, and both making the tool adoptable, and doing the adopting will take a lot of people’s time. While not as far along, helping Red Hat’s konflux team work in Fedora, and build it into an essential tool we’re all delighted by is a good second example of what I have in mind. A separate post on this topic is in my queue, actually.

Completely agree it should be open and transparent, and that what to cover is a big open question. The way I look at it you need to start with something focused enough you can do it justice with the time you have, and distro/community-crossing enough to be relevant. To my mind this suggests topics related to distribution fundamentals, the combined whole, the software that runs it, the timetables between them, and so forth. I think the only way we really find out though is to have the conversation, try it out, make some changes, try again- basic iteration.

I would begin with a big messy conversation with members of Fedora Council, the CentOS Board, and some of the distribution leads in RHEL. This wouldn’t be The Thing, it would be the conversation that helps us figure out what we all want The Thing to be, basically the answer to your question. I can organize this if the Fedora Council is interested.

1 Like
4

This sounds like a smart starting point. A couple questions that come up in my mind are:

  1. Are there previous and similar (or similar enough) attempts to bring people together that we can learn from?
  2. How will we know if this has been successful? It can be tough to bring groups together and even tougher to get sustained focus and effort in today’s fast paced, multi goal’d, and limited time world.
5

I want AI. :grin:

6

That’s weird, my Mastodon feed is full of people saying the exact opposite :wink:

Seriously though, can you narrow down what aspects you’re interested in? Some people are looking for the software stack for their own use, others are looking for models, others documentation for either, and of course others still are looking for AI capabilities inside Fedora infrastructure. And probably so much more. We’re working to get a small amount of equipment with GPUs in our data center for community infrastructure use. Some of that is about breaking the chicken-or-the-egg problem we had last year.

7

Probably? My context is inside Red Hat where my standard move is: Put people together, talk openly, discover needs, make goals, execute. The dynamics in a 3 way community space may differ, having elected or nominated leads representing constituencies may suggest particular practices. Perhaps others with more community building experience under their belt can comment.

That’s a great question. I have a tiered personal answer, but your mileage may vary. For me initial success would be that conversations inside one community that have potential impact on the others would make their way into those affected communities rapidly and without having to resort to special measures (Like planning a talk at a conference). Further success would be bringing decisions in one community into the whole, at least for discussion’s sake. I think a lot of the time we hold back sharing incomplete ideas because we don’t want to appear ill-considered or message a “maybe”, but the reality is when we finish something before it is ever shared all we’ve really done is denied people the option to take part of its creation.

1 Like
8

Sometimes creating something without every-bodies input is much more efficient. Linux, despite its fragmentation, works so very well and moves so very fast. I know that my input to technical questions of e.g. a new git system would just slow everything down.
In some other inter-departmental and inter-agency agendas, important decisions are made over a cuppa by very few people. This allows for lean and meritorious decision-making.
What many people want is their egos massaged and to be ‘included’. Fran Lebowitz says it well, but rudely. Not everyone deservers to write their book or tell their story.
‘What everyone wants’ should definitely be part of extreme-long-term planning. But what works for technological development of Linux at this time? It is a competing fragmented landscape of interests and abilities in which excellent ideas quickly rise to the top.

9

AI that helps with tasks. LocalLLama. Downloading models, sharing saving memory. Decent UI like the login screen could be a task prompt “What do you want to do today?” and then your “window manager” is basically Roo Cline. The machine can be controlled remotely acting as my AI agent.

1 Like
10

I agree with @abitrolly that improving the integration of AI in Linux is important, as long as it doesn’t rely (…too much?) on cloud services.

Another thing I think could be improved upon would be to improve the overall gaming experience in Fedora/CentOS/… in some way shape or form. Though I’m not sure what kind of effort is needed in that regard.

1 Like
11

Yes, this isn’t a one-size-fits-all approach, and often you want to make a little headway in doing something new before opening it up. But if you take it all the way to the finish line, it might only be good for what you’re interested in, and be a miss for everybody else. It’s not that there’s only one correct answer, it’s that you’re making a tradeoff.

My core premise is that the tradeoff we have right now, information sharing between the 3 distributions, could benefit from exchanging information and soliciting feedback earlier.

12

That is majority of what I don’t want in a desktop OS :stuck_out_tongue:

I might accept AI at the desktop if it was 100% all-local with everything involved (including the hardware/NPU). I’m not into introducing erratic dynamic cloud behavior to my local OS.

Generally speaking I love the idea of telling my computer to do stuff and it happen; I just don’t believe it can happen at the speed and consistency I want considering I have the two top AI consumer speakers in the US and still have to give multiple prompts to play music occasionally; along with them randomly initiating listening based on TV commercial noise. Aiming a mouse cursor is less potential nonsense :stuck_out_tongue:

13

A noble and useful goal. Perhaps a Red Hat funded meet-up of the core people twice yearly, and a long-range planning forum online for all users?

And no, I don’t want AI integrated in my daily driver. There are many programs and organizations empowering users to run and customize their own AI systems. What I want is enhanced and consolidated publically available documentation across the RPM system.

3 Likes
14

I think we already have enough leadership meetings. When Red Hat wants something, they should show up in existing community spaces to talk about it. But, I like the core idea of more communication.

What about, instead of an “interlock” meeting, we did a monthly podcast[1] with a panel of people in Project and Red Hat leadership. Each session could have a theme and some pre-seeded topics, but it’d also have a live “call-in” panel Q&A.

I’d also like us to think about extending the panel (at least, in invitation!) beyond RH / CentOS / Fedora into other direct Fedora downstreams — Amazon Linux, Universal Blue, Asahi (depending on how we’re counting), and even smaller ones.

  1. or, vodcast, as it may be ↩︎

3 Likes
15

In 2025, I was looking for new ways to support @itguyeric and @joseph on the Fedora Podcast. I hoped to leverage an outside editing group to help provide editing support and show notes. Maybe we could make this into a recurring segment on the existing Fedora Podcast, or maybe we should fork it off as something new?

Open to feedback on what folks think is best. I am actually in favor of a conversational format for discussing project “baseball”. To me, it is more natural and the editors can make the conversation into easy listening by removing the “umm” “ahh” and other pauses in the conversation.

16

We could always add an extra episode per month to the podcast, every other week + 1 sort of thing, just as long as the format didn’t stray too far from what folks hear already.

1 Like
17

Maybe some of them could be used for this thing at certain intervals.

Thing is, I’m not proposing a “what red hat wants” thing. I’m proposing a “what we’re working on that you might be interested in” thing. That’s a 3-way thing. Of those 3, I suspect people will be most interested in what RHEL is up to since we’ve been terribly opaque. On the other hand, many activities in Fedora lately have been worth talking about :slight_smile:

That definitely has potential. The goal is the mutual sharing, the visibility, taking advantage of our commonality.

My thought was let’s start by getting leadership from the 3 bodies together to see what is the stuff of mutual interest. If these things are, let’s find a way. If they’re not, it might just mean “not yet”, and so we start from what everyone wants, and grow from there.

18

One of my favourite sayings from philosophy is Occam’s Razor, which says “take the simplest route”, or my preferred formulation “don’t multiply entities beyond necessity”.

The most difficult thing interacting with the Fedora ecosystem is the multitude of semi-authoritative sources available. Take for example the Fedora Magazine. It is a cool idea, yet articles are rarely published, and I am yet to find tutorial that works in there. I would remove and consolidate properties.

For a new media show to be sucessful, there would have to be a paid journalist working one or two days a week on it. And then how would it be able to be discovered?

Could I suggest instead that for all media, existing properties be utilized? These discussion forums are the place for that.

The problem is not one of the lack of a ‘forum’. The problem is of siloed workspaces and differing management expectations. Fedora users and packagers and active on this discussion forum, yet technical leads are on Bugzilla. I occasionally see brief marketting efforts from Red Hat here. I never see CentOS users here.

To solve the siloed workspaces issue takes different things for different workspaces. At corporate Red Hat, it takes management to direct paid resources (time) to interact with this fora. Fedora users are here already. Cent OS I do not really know about so perhaps some else could suggest what motivates them.

Fedora’s greatness is in its ability to move quickly and remain at the cutting edge. It is a lean machine! It is a friendly and responsive community. Red Hat’s strength is its solidity and certainty.

19

Or a livestream? With video/podcast replay of course to accomodate for timezones. The livestream’s interactive format is very appealing, IMHO.

1 Like
20

I think this hits one of the things I’m worried about with this. There are many ways to see what’s going on in Fedora (and CentOS). Red Hat (in general, outside of those of us working on Fedora directly) doesn’t follow these as much as I’d like. I’m worried that an official interlock meeting might exacerbate that. (“We’ve got that covered — there’s a meeting!”)

Well, we have the Fedora Podcast, all done by volunteers. So, don’t sell that short. But, also… Red Hat does have people who work on this kind of thing… @blc, maybe we could get some support for this if we put together a good case?

Yeah, that’s exactly the format I was thinking of.