= phenomenon =
Following the workstation product discussion, it seems possible, if not likely, that the Workstation will not include all desktops that currently exist as spins. Those that are not included may want to have a product similar to Workstation.
Similarly, there may be new areas of development in Fedora that arise that target new areas, new use cases, and might want to have a new product.
How should these ideas be submitted? How should they be approved? etc.
I propose we defer creating new products for one release so that we can work out some of the kinks and identify some of the areas that require additional man-hours whenever we add a new product.
That said, do we want to still talk about procedure for adding or do we want to leave that to a future fesco?
I would like to discuss it because, given the Spins and SIGs that exist in Fedora now, we should have an answer for how they can accomplish what they do now in a Fedora.next world, whether that's a new spin process, a new product they maintain, or something else. I don't think "come back after F21" is a very nice answer to give them.
We can possibly postpone discussion until we know how we're building the existing products out of the Fedora repo(s), but I think we need an answer before F21.
Should this be a meeting item now, or after we tackle the PRD submissions?
Couldn't be tweaked spins process used for products? With multiple tiers/classes - aka tier 1 product will be release blocking, marketed, part of Ambassadors DVD etc. Tier X non release blocking, with marketing coverage etc. Of course, with PRD and resources commitment requirements.
But the overall aim should be, that Fedora stays inclusive community project.
ACTION: mattdm will create proposal for spins/secondary products (mmaslano, 18:12:00) ACTION: jreznik will help mattdm wiht proposal (invite interested people...) (mmaslano, 18:15:44)
mattdm continues to be aware of this on his todo list. :)
Deferring some more to give mattdm time to write up a proposal.
Link to thread https://lists.fedoraproject.org/pipermail/devel/2014-January/194877.html
Unfortunately the thread spilled over into February so you'll have to hunt around by subject in the next month's archives as well:
https://lists.fedoraproject.org/pipermail/devel/2014-February/thread.html
I still have the ideas I was talking about before in mind, although I was kind of letting the spins discussion play out, and really the way that went has made me think I need to refine the idea.
My previous thought was that all spins would become secondary products; now, I think many spins should just stay spins and that some may wish to become "incubating products". These would
I don't want to premature hardcode the exact requirements to be a primary product, although maybe there's room to be a little less vague than I'm being here. :)
If we do end up with a number > 3, it might be good to have periodic (annual?) board review of the success and continued strategic importance of each (including the initial three we've already approved).
Replying to [comment:12 mattdm]:
Aside from the criteria you have here to help the spins/products focus, we already do (did?) all of this. When I was on the board the first time, we did the whole "evaluate what to promote on the website". This is essentially what you're proposing again, unless I'm completely misunderstanding what you're saying.
I'm not convinced at all that setting up an incubator process is going to actually result in any meaningful benefit. Either the spins wind up working through the process and getting a rubber stamp to be a product which really leads to confusion on the download site, or they work through the process and wind up being rejected because of strategic importance and then getting discouraged. Also, doing the review and then demoting a product is likely to be a pretty unpopular and difficult to accomplish task. I do not think we can engineer process to get rid of the fact that there are primary and secondary things within Fedora.
I personally think Spins, other than DE spins (MATE, Cinnamon, etc), really suffer from the fact that they have no alternative deliverable form other than an iso. Which means they suffer the burden of being a full "product" that has to get composed and tested when all they really need to do is deliver a set of packages pre-installed. Figuring out a model where non-DE spins (or even those in the case where the DE is parallel installable) are able to leverage the vast majority of the work the products are doing is probably going to be more beneficial in the long run.
(Sorry, not on FESCo)
Replying to [comment:13 jwboyer]:
I'm not convinced at all that setting up an incubator process is going to actually result in any meaningful benefit. Either the spins wind up working through the process and getting a rubber stamp to be a product which really leads to confusion on the download site, or they work through the process and wind up being rejected because of strategic importance and then getting discouraged.
The requirement that the proposed product have a clear problem space / ecosystem fit approved at the very beginning by the board is meant to address both a) the issue of what to promote to whom in what ways and b) someone putting in a lot of work and having it rejected. Do you think we could make that stronger and have it address the problem better, or do you think that the problem is more fundamental than that, and if so, is there another solution?
Also, doing the review and then demoting a product is likely to be a pretty unpopular and difficult to accomplish task. I do not think we can engineer process to get rid of the fact that there are primary and secondary things within Fedora.
@sgallagh, you've mentioned several times that "fail fast" is an important idea of your conception of the three products proposal. Want to weigh in?
I agree with this. I think the ideal situation is that DE spins are supported by the Workstation WG, and non-DE spins transformed into something other than an install-time solution.
It would also be nice if we could also use the word "spins" for different flavors of the cloud image -- I keep doing that accidentally.
For new products, I'm thinking more about Fedora for Phones or something big like that -- not that I think that's a good idea right now, but... something that fits in a very different space. I could also be convinced that it wouldn't be ''completely'' terrible to have different desktop-oriented products with some overlap, if someone really demonstrates that they can put up the work. On the other hand, I know that some people ''do'' think that ''would'' be terrible. And on the other other hand, I know that some people think that it'd be ''ideal''. Imma go get some coffee now.
Replying to [comment:14 mattdm]:
Replying to [comment:13 jwboyer]: I'm not convinced at all that setting up an incubator process is going to actually result in any meaningful benefit. Either the spins wind up working through the process and getting a rubber stamp to be a product which really leads to confusion on the download site, or they work through the process and wind up being rejected because of strategic importance and then getting discouraged. The requirement that the proposed product have a clear problem space / ecosystem fit approved at the very beginning by the board is meant to address both a) the issue of what to promote to whom in what ways and b) someone putting in a lot of work and having it rejected. Do you think we could make that stronger and have it address the problem better, or do you think that the problem is more fundamental than that, and if so, is there another solution?
I think it sets the Fedora Project governing bodies up for unrealistic expectations. Nobody, including the Board, likes telling people the thing their proposing isn't important enough to do because clearly someone sees a need not being filled. In the absence of something else filling that gap, I can't see the Board or FESCo denying an incubation request. Really though, if the problem is marketing to those niches, then I think we need to figure out how to adapt marketing and tooling of existing products to reach those people rather than just approve new products and hope that's enough.
Now, clearly that won't work for everything and I think there may be some things that really require a new product. However, I think the bar should be incredibly high for a product approval. Otherwise I think we'll wind up with a product state where everyone is special and they all get the same medal for having played (please forgive the sports analogy). The steps you have for the process seem fine to me, but I think there should be very clear messaging that the default promotion answer will likely be "no", and that working with existing products is a better way to begin.
Also, doing the review and then demoting a product is likely to be a pretty unpopular and difficult to accomplish task. I do not think we can engineer process to get rid of the fact that there are primary and secondary things within Fedora. @sgallagh, you've mentioned several times that "fail fast" is an important idea of your conception of the three products proposal. Want to weigh in? I personally think Spins, other than DE spins (MATE, Cinnamon, etc), really suffer from the fact that they have no alternative deliverable form other than an iso. Which means they suffer the burden of being a full "product" that has to get composed and tested when all they really need to do is deliver a set of packages pre-installed. Figuring out a model where non-DE spins (or even those in the case where the DE is parallel installable) are able to leverage the vast majority of the work the products are doing is probably going to be more beneficial in the long run. I agree with this. I think the ideal situation is that DE spins are supported by the Workstation WG, and non-DE spins transformed into something other than an install-time solution.
I'm going to assume by "...supported by the Workstation WG..." you mean leveraging the Workstation product. If so, then yes!
It would also be nice if we could also use the word "spins" for different flavors of the cloud image -- I keep doing that accidentally. For new products, I'm thinking more about Fedora for Phones or something big like that -- not that I think that's a good idea right now, but... something that fits in a very different space. I could also be convinced that it wouldn't be ''completely'' terrible to have different desktop-oriented products with some overlap, if someone really demonstrates that they can put up the work. On the other hand, I know that some people ''do'' think that ''would'' be terrible. And on the other other hand, I know that some people think that it'd be ''ideal''. Imma go get some coffee now.
Fedora for Phones would be one of the cases I could see needing an actual product. It's somewhat of a misleading example though, since the rest of the products are addressing using general purpose hardware whereas phones are pretty darn specific hardware and use cases. However, something like "Fedora Big Data" is what I would consider to be a layer/role (maybe we use layers instead of spins?) on top of e.g. Server.
So perhaps you and I aren't really disagreeing on the concepts, but I think we really need to address the expectations before we even get into how to be a new product.
Replying to [comment:15 jwboyer]:
This seems good to me, but I'd also like to hear more from the opposing view (low-bar would be better), and while I don't think we can really reconcile the two sides, at least get a general understanding that we've made a community decision together. In other words, I frustratingly vote for Further Discussion.
However, ''I'm'' not going to keep playing devil's advocate for the low-bar-many-products-is-fine approach. I think you've put the high-bar reasoning fairly clearly; at least, I think I understand it. Is there anyone on FESCo interested in advocating the low-bar side? Or not on FESCo -- Eric Christensen, maybe? I know that Jóhann Guðmundsson has written passionately about this (arguing, I think it's fair to say, for Fedora to stay only in the business of producing toy bricks and to stay out of offering them as sets), but I'd like to hear from some other people as well -- particularly, I'm pretty sold on the idea that we ''should'' offer sets rather than just bricks, so I'd be interested specifically in arguments for doing that but not limiting their number. (Along with some practical ideas for how that would look on the web site and in marketing materials.)
If it's pretty clear that most people willing to step up and talk about it are tilted to being focused on fewer products, that makes it easy. And I can stop being frustrating. :)
I agree with this. I think the ideal situation is that DE spins are supported by the Workstation WG, and non-DE spins transformed into something other than an install-time solution. I'm going to assume by "...supported by the Workstation WG..." you mean leveraging the Workstation product. If so, then yes!
Yes.
In fact, we've proposed such a... spin... for Cloud.
Also, I just want to add these two blog posts. They happen to be from Fedora board members, but that's not the particular point. And they're from kind of near the beginning of all this, but I don't think that matters either. I think they represent both of the camps pretty well:
I mostly agree with Sparks with some pieces of mjg thrown in. I see it this way:
Fedora the Project should follow Sparks' vision. But in order to produce compelling products, we need to let the Products themselves specialize and focus. What does this mean? The Fedora Project needs to work towards a larger set of products which each suit some people. The individual Fedora Products need to be able to make decisions that may conflict with other products and we need to figure out how best to deal with that.
For instance, if we think that one web page with 20 Products to download is a mess then we need to look at solving that by driving people to web pages for the individual products rather than the overall product web page. The Apache Foundation's incubator was cited by other fesco people as where they want to position the Fedora Project in the future. I would point out that http://incubator.apache.org/ is a list of 36 Products. But people who are using those products aren't going to incubator.apache.org to find them. They go to the products' home page (example, http://cassandra.incubator.apache.org )
If we attempt to lock down too much how many Products there are then we're just going to create an artificial scarcity of Products. Just as with F < 20 we would periodically see proposals to replace GNOME another DE we would see proposals to replace Gnome in the Workstation Product with an alternate DE, GDM with an alternate DM, alternate screensavers, terminal emulators, etc. We need flexibility at the Project level to enable focus at the Product level.
What Toshio's comment brings to mind is whether we actually want to use Fedora as the brand for Fedora products; it becomes a more clear exercise in branding and marketing if they are using some per-product brand. That certainly does have other downsides, though.
I think my representation of this as high-bar vs. low-bar isn't quite right. That's certainly one possible axis, but the other one is product-story/market-need-based vs. demonstration-of-resources-based. I think consensus is generally that we want a high bar for Products, but it's less clear about what type of bar we actually want.
Log in to comment on this ticket.