Josh Saddler

Now that the latest stir in the gentoo-dev ML has been slashdotted, I thought I'd speak out as a voice of reason.

First, the Seeds project hasn't even decided exactly what they will release (though a canned LAMP stack is a likely thing), nor in what form. They've said they plan to use catalyst to build these stage4 tarballs, but catalyst might not be the only thing they use. They have very few rules in place about what they'll do.

Second, they wish to remain a top-level project, running autonomously, under their own auspices, etc. Not as a subproject for any other group.

Third, the devmanual and other rules say that project creation is allowed, even competing projects are allowed, all without creating GLEPs to approve them.

These points being made . . . several people simply don't get it. Some members of releng in particular feel slighted, that Seeds encroaches on their turf, that it's created more work for them, and so on. All the arguments against Seeds are based on one or more incorrect views of the above three points, whether ignored, not understood, or just forgotten. Let's take the most common arguments thus far. Note that I've personally talked with most of the parties who created these, so I'll use a mixture of seriousness and tongue-in-cheek.

- "Stop giving releng more work!" aka the releng refrain
No one is giving releng more work. You assume the onus is on you, when in fact it's the Seeds folks who create their own work. Let's say that Seeds gets around to actually releasing stages, rather than just working with profiles or specfiles (as previously touched upon in the thread). All that needs to be done is creating a new Bugzilla component to properly handle the Seeds stages/releases. I think proper Bugzie handling would go a long way toward ensuring proper workload distribution. Certainly no one is saying that Seeds requires releng to build and test their stages/specfiles/profiles/whatever. There's nothing that says Seeds stages will even be simultaneously released with 200X.x releases.

- "It needs a GLEP!" aka the ciaranm whine
No it doesn't. Period. Nothing in our rules says project creation must have a GLEP. GLEPs exist for "wide-ranging feature or enhancement" for things that affect Gentoo, e.g. Portage tree additions, new package managers, new ways of committing, and so on. As I said on the lists, there's a difference between a project that can be widely used by users and devs, and one that forces wide-ranging changes on everyone. Potential != forced effects.

- "He said this! Sneaky! Didn't get approval from so and so!" aka the underhanded allegation situation
This is actually a combined subset of the first two arguments. First, the Seeds people don't need permission from anyone -- not releng, and not from an approved GLEP. They're free to create what they wish, and to structure themselves with the same freedom. There was some early confusion in the first mail -- it seems that Stuart said he talked to some releng folks, who in turn have no knowledge of this, and that understandably annoyed them (some to more profane degrees than others, of which I have interesting logs), but really, they don't need anyone's permission to go about their business, as their efforts are not GLEP-worthy; i.e. do not directly affect all the other projects or Gentoo as a whole -- we can use their stuff. We're not forced to. I've heard them called "sneaky and underhanded" by agaffney, but there is no way that was their actual intent. Probably would have been better if Stuart simply hadn't mentioned releng at all.

- "Not enough cooperation!" aka duplicated work worries
This is actually a semi-valid point. kloeri brought it up on IRC, and while it's always nice to think that projects aren't competitive, that they'll contribute to each other (especially if they use the same tools e.g. catalyst), the reality is that we have any number of projects that don't have any particular level of cooperation or interaction with each other, whether good or bad. And again, until we all know more about what they plan to do, there's not much point in planning for or demanding levels of cooperation, or who answers to whom between projects. In response to this point, however, both releng and Seeds agreed to a liaison, so there's some progress right there.

- "You must implement feature X!" aka wanna-implement-these-old-GLEPs?
Carlo brought this up by demanding that Seeds implement user-specified Portage sets. Turns out that's not really the focus of the project. While the Seeds people are looking for ideas, it's not polite to order them to do something, is it? Be civilized; pitch your ideas to the appropriate people for discussion, but don't demand that something be added to a project not your own. You know how proper feature and implementation is supposed to go. (At least, the devs are supposed to.) Portage sets? So now Seeds is doing work for another project -- optional Portage work, at that? Please.

I'm sure there will be more arguments in the coming days. But as I said on IRC and the lists, new project, fresh start. Wait to see what they come up with before making judgments. This kind of instant, inflamed negative response to project creation is discouraging at best. I don't want to see another Sunrise debacle -- we lost a few good devs (such as brix) because of that.