On Sun, 2022-03-27 at 18:05 +0200, Jonathan Carter wrote: > Is that the kind of indifference you're referring to?
There are lots of situations where indifference is a factor; the factors you mention, derivatives that no-one in Debian ever gets to hear about, long standing derivatives that are well known to Debian contributors, derivatives for particular niches or for specific parts of the world etc. > Or do we need better interfaces with them all together? I think we need better relationships with and connections to the world outside Debian, including the world of derivatives. > I'm guessing that you're speaking about derivatives like Ubuntu and > Devuan. I think enough people are familiar with the complexities behind > those so I won't go into them now, but are there any specific animosity > with a derivative that we could've avoided that I might not know about? Correct. None that I can think of right now. > we list Ubuntu information in our QA pages (like which version they > carry, whether there's bugs filed, patches, etc). Part of the reason for that is that other derivatives are often not as well organised that any of this is possible at all. Personally I would like to see our QA pages treat derivatives more neutrally, and the derivatives census was part of that. > Do you think the derivatives team could do something like host a video > call, inviting all derivatives who would like to submit feedback about > every 6 months or so? Then at least some more people from both sides > could get to know the people on the other side, and likely some problems > can get solved too. At this point in time, the derivatives team doesn't really exist, so that seems unlikely. When DebConf was in person, it fulfilled the role of connecting people for some derivatives and I think that it is better to have connections between derivatives and the wider Debian community than just between derivatives and a Debian derivs team. > Well, I'll admit that I'm a bit ignorant on our biggest problems with > our relationships with our derivatives. > > Can I through this question back to you and ask you to share some of > your insight and concerns? There are a mix of improving things and worsening things, for example: We still have new Debian members coming from derivatives (congrats Nick Black!). Lots of folks from derivatives present at or attend DebConf. We have seen some derivatives like Deepin work on merging their independent desktop implementations into Debian. Devuan folks are contributing positively to Debian these days. Some derivatives contributors, including from our closest derivatives, and including Debian members, don't make changes for issues also affecting Debian directly in Debian or even communicate back fixes to package maintainers. This probably always happened but I noticed it a bit recently. > I suppose our approach to derivatives is quite similar to our users, we > through all these source and binary packages and isos out there and then > everyone gets to fend for themselves. Sure, we do have documentation and > some support channels, but for the most part there's not lots of hand > holding involved. Agreed. > Also, do you think that's a problem? I think it leads to various probably sub-optimal situations; for eg duplication of work, distro proliferation, stagnation in Debian, users going elsewhere for their software needs etc. > One improvement I'd like to see is PPAs. As I understood it, Debian bikesheds were not for organisations outside of Debian to use, just for packages not fitting into the current set of defined suites. > It would also be nice to offer some infrastructure to build their > installation media for them. All of this might make it easier for them > to get started, and also, perhaps create a smaller delta towards > becoming a pure blend if it makes sense to do so. I think we don't even have this or live images for Debian blends yet, but many derivatives have their own live image setup (forked versions of old live-build releases for example) and many have their own independent or cross-distro installer implementation (Calamares etc). > Not sure if you're familiar with extrepo? As I understand it, extrepo is more for things like the Mozilla Firefox or PostgreSQL repositories than things like Ubuntu? Probably a discussion for the extrepo maintainer, or potentially there is room for an extrepo extension containing apt configs/keys for derivatives, perhaps pulled from the census. As an example of where this would be useful, I'm pulling the Ubuntu census apt config into a chdist so I can list Ubuntu package versions etc on the command-line. > > What is your favourite derivative? > > I'll go with Kali Linux, I think they're a good example of a good > derivative. > > > Would you like to see it merged into Debian? > > I'm quite ambivalent really ... especially if it's going to cause > some friction. As Kali is somewhat of a commercial distribution (I feel like it is a marketing advantage for its corporate sponsor), having the Kali brand go away is probably not going to happen by merging Kali with Debian. They are already working on merging their packages into Debian though. https://wiki.debian.org/Teams/pkg-security > Plus, right now we're in trouble with our live build infrastructure, so > currently for testing we're not even building our own live images, it's > pretty much where my technical work will be focused on in the immediate > future, and I hope that we'll have non-x86 live images for bookwork and > also a live image for debian-jr, but now I'm getting a bit side-tracked :) It is great that you are considering reviving Debian Live (I was expecting Debian bookworm would have no official live images), but I expect that most of the Debian community don't know that there is a problem and that there needs to be work done on Debian Live, so please consider a d-d-a post about this issue, other folks might contribute. > > Thoughts on identical Debian pure blends vs derivatives? > > Not sure I understand the question? I'm alluding to the advantages of having an independent identity etc. As an example, consider Kali vs a Debian Pentesting blend. Kali is a well known brand in the security auditing industry, and even outside of it. I expect that a Debian Pentesting blend would not be as popular. OTOH there are Debian pure blends with a separate identity; FreedomBox. The other aspect is "download and boot Kali" vs "install Debian, then install this metapackage via the command-line". > Are there pure blends and derivatives that are identical to each > other? There are certainly similar ones; for eg multimedia and Ubuntu Studio, but the situation in the question was mostly hypothetical. > Well, we've talked about this a few times over the years. In 2012 in > Nicaragua you first introduced me to the derivatives team, and the > derivatives census. I never contributed to it much but I still think > it's important work. I've been on the distro side to and found it > valuable because the census also checks that some things in the distro > is configured correctly for that derivative (like the origins data). Its nice to hear that derivatives folks liked it. > Last I spoke to you, I think you also said that you don't have much time > to work on this anymore, and that the distro census is probably going to > die down. Is this still the status? Do you think it can be saved? We > probably have quite a wide audience here, do you think it's worth > another shot to get people involved to work on it? The census has been turned off for years. It is too much work for one person to do on their own. I don't have motivation or time or ability to do it alone any more. I have tried to recruit other folks to work on both the social and technical sides of it. I had an Outreachy intern that did some great work. I had some interest in contributing from both DDs and non-DDs but the interest didn't result in the sort of ongoing contributions that are needed. I've had encouragement for keeping the census around from a couple of folks though. I'm also not sure the Debian community thinks the approach is correct or even useful to Debian itself and also for derivatives themselves; the mailing list and IRC channel are mostly silent for years. There are also some solvable technical flaws that mean the census cron can't be turned back on right now, and the motivation issues block fixing them. -- bye, pabs https://wiki.debian.org/PaulWise
signature.asc
Description: This is a digitally signed message part
