Hiya, [ ccing debian-derivatives: This is a thread[-1] which started on ubuntu-devel, and which I've turned into a thread about uploading new packages to Debian instead of Ubuntu directly. It's not really about Canonical software, more general packages. Input is sought about your POV on this. Specifically whether you think, in general, that us redirecting contributors from REVU (our mentors equivalent) to Debian is something that should happen/continue. ] Replying to both mails in one. On Mon, Aug 02, 2010 at 02:49:10PM -0500, Micah Gersten wrote:
On 08/02/2010 02:35 PM, Paul Sladen wrote:On Mon, 2 Aug 2010, Mackenzie Morgan wrote:pushing more for the packages to go to Debian first and then syncTurn-around times have been one suggestion /why/ pushing to Debian in the first instance does not happen enough---it is a lot easier to manipulate an upload directly than wait for it to be proxied via a third-party (but with REVU it could be argued that there is already that third-party). To short circuit the difference in delay period between 'dput ubuntu' and 'dput debian' + sync back would effectively need a skilled DD able to act as that proxy for new packages; and most of the people of sufficient calibre are probably already working on their personal packages, or $dayjob packages. It's likely to be mind-numbing, which probably means bribing such a willing individual.
As we are talking about individuals who do not have archive access themselves, and given that the delays for REVU processing are astronomical (tending towards infinite?), I don't think we need to be concerned about increasing the turnaround time. I don't see why it should be one single individual who does all of this sponsoring. It only seems fair to Debian — given that we are proposing to put stuff in their archive too — that we ask our contributors to integrate properly, and that means breaking out into the wider Debian world. Most DDs are willing to perform sponsoring and, even better than that, there are packaging teams[0] which I've found to be very welcoming to new contributors.
That one individual would be ultimately signing their name on lots of packages, diluting their reputation if long-term maintenance doesn't end up being forthcoming after they're marshalled the initial uploads. Back when people like Scott and mjg59 were still DDs I found it relatively easy (and therefore not overly onorous) to get uploads done on a semi-predictable turn-around.
Actually, in addition to what I've said above, we now have the #debian-ubuntu channel on OFTC set up. I believe that one function of this channel could be to facilitate sponsorship in Debian. There's also the Derivatives Front Desk[1], which is part of the same initiative.
The question is, how to return to that situation of a semi-reliable 24-hour turn-around without forcing everyone through the Debian New Maintainer process in parallel. Presumably the reason people are following the /Ubuntu/ path in the first place is because of a perception of an easier welcome with gradulated steps to direct involvement.
There are many great things that MOTU hopefuls and MOTUs can do, but I don't think that (in general), maintaining entirely new packages has proved to be one of these things.
-PaulAnother side of that argument is, do we really want to take in a lot of packages without that maintenance commitment? The nice thing of pushing through Debian is that someone is committing to maintaining the package. Also, I think bdrung or someone said in -motu that make sure the packaging is up to standards and then push through Debian. Without the actual commitment of maintenance, MOTU ends up with a lot more work to do. Maybe we should find a way to get more MOTUs in the position to sponsor uploads in Debian as well?
I agree. Historically we have been really quite bad at maintaining packages in Ubuntu: experience shows that people lose interest after a short while and stop caring for uploads. There are no mechanisms to catch this, and with our team development model there's really no way to exert pressure on a particular individual to perform work anyway. It's not clear if having ones name on a package in Debian will make this situation any better. One of my Big Things is contributing to Debian directly instead of making uploads to Ubuntu. I think that MOTU functions best when it performs a QA role, and that everything is so much smoother when work is done as far upstream as possible. Most packages — especially ones that turn up on REVU — will work on both distributions using exactly the same source package. To bring this mail back to the original post in the thread, I don't think that REVU days or a REVU sprint are solutions. I think that reducing the use of REVU is. You may argue, fairly, that this increases the barrier for new contributors. But that's OK. Maintaining a package is hard work. It requires commitment and graft from one or more people. I believe that each package needs someone (be that an individual or specialised team) caring for it. I believe that our model doesn't provide for this. I further believe that, philosophically and relating to being good FOSS citizens, we should be doing our packaging work where most people can benefit: in Debian. There is room for us to be flexible here, too, when the situation demands it. People are making an incredible contribution by developing new packages for us. We should be careful not to demotivate them by making this feel like jumping through hoops. It really is the right thing to do, benefits more people and will actually lead to better quality software packages in the end (in the main, you will find your specialised Debian sponsor to be more knowledgeable about your particular domain than a random MOTU). Sorry for turning this into a soapbox-style diatribe. :) Ta, Iain [-1] https://lists.ubuntu.com/archives/ubuntu-devel/2010-August/031034.html [0] http://wiki.debian.org/Teams#Packagingteams [1] http://www.debian.org/News/2010/20100629
Attachment:
signature.asc
Description: Digital signature