I went across statistics from Fedora Package Database and what caught my attention is that the increase of number of packages in the official Fedora repository has almost stalled:
The number of packages in Fedora 22 is 17021 and is not going much since Fedora 20. Does it mean there are no more projects worth packaging? I don’t think so. The number of open source projects goes up like never before, just look at GitHub.
I think this trend is related to the growth of Copr. The number of projects has been rising exponentially there. Mirek Suchý reported a couple of months ago that the number of projects in Copr was almost 3000 and almost 2000 were active. And the numbers have increased significantly since then.
It’s actually a success. It means we have achieved what we outlined in the Fedora.Next plans: we’ve built a ring of software around Fedora which has low barriers to entry for packagers and where software is easy to install for users. Although the number of packages in the official repositories is not growing like in the past the total amount of software available to Fedora users has grown like never before. That’s great.
What we’re still failing at a bit is how to build on this and bring the best of Copr to official Fedora repositories and convert the most promising Copr packagers into Fedora packagers. The official repositories still have their relevance. The quality of packages there is significantly higher than in Copr. We should encourage Copr packagers to work on their packages to meet Fedora standards and become Fedora packagers. We should show them the path. I can imagine that we offer an option in Copr to run the source packages against fedora-review to give the packager a hint what needs to be done to meet the official repository standards and if he/she is interested we can point him/her to documentation for the rest of the process.
The current situation is a great opportunity if we streamline the path to quality. Then Copr can serve as a broad source of “playground” software from which useful and proven projects can get deserved quality of integration and make it to the official repositories. But it’s also a threat because if we don’t provide a path and encourage Copr packagers they may just be satisfied with the easy way to make and maintain packages in Copr and no one will want to package software for the official repositories any more.
Brno hat 
It would also be interesting to know the percentage of COPR repos that contain packages that are / are not already packaged in Fedora
All my COPR repos are just newer versions / master versions of packages that already exist in Fedora.
There are definitely quite a few projects that just include newer versions of packages, but still I think there is a growing number of repositories which include software that is not in Fedora repositories. People just package it to the point it’s working and leave it there. Take me as an example. I packaged one Pidgin plugin there, it built, it’s working and I have never found motivation to work on the package to get it on the level of official repositories. I’ve never been encouraged and I don’t even have a clue how difficult it’d be, maybe it’d take just a few changes.
I finally made my first RPM for COPR:
https://ttboj.wordpress.com/2015/07/08/oh-my-vagrant-mainstream-mode-and-copr-rpms/
Making the RPM was still the most painful part, and I’m not entirely convinced I didn’t break some Fedora RPM rule…
So making an “RPM version next” would be really awesome…
As a second aside, I noticed the copr plugins package that is needed to use the dnf copr enable command isn’t a default in fedora. If we care about COPR, I think it should be 🙂
Cheers!
James
A path from Copr to the Fedora Package Collection? Via the Package Review process? Suddenly, packages from Copr won’t be accepted, because they don’t adhere to the packaging guidelines or need lots of more work (bundled stuff, licensing issues, conflicts, …).
Are there any data/statistics about packages in Copr that have been submitted to the package review queue? http://fedoraproject.org/PackageReviewStatus/
How popular are Copr repos? What are the most promising packages in Copr? Who are the most promising Copr packagers, who are not in the packager group yet?
“Suddenly, packages from Copr won’t be accepted, because they don’t adhere to the packaging guidelines or need lots of more work (bundled stuff, licensing issues, conflicts, …).”
Sure, that’s what I’m talking about. But we should motivate packagers to go through it otherwise we’ll end up with just low-quality packages in Copr. There won’t be any additional licensing issues because Copr has the same license requirements and for same simple packages there actually might not be so much additional work. Right now, you never know until you decide to dig into it. We should give them some hint because if they see it may not be so much work, it may be encouraging.
Your questions are good, but I don’t know answers to them, I don’t have access to Copr data beyond what’s publicly available.
I am running a COPR exactly for this reason. Ok, the package is pretty large and the codebase is not “pretty” but I opened the review request 3+ years ago… https://bugzilla.redhat.com/show_bug.cgi?id=804125
@Gianluca : Consider swapping reviews with somebody. That works for in many cases, but requires you to be a more active reviewer, too, instead of waiting for others to run the show. The ticket you’ve linked is assigned to a reviewer since early 2014, which hides it from the normal review queue: http://fedoraproject.org/PackageReviewStatus/ If there’s no progress after such a long time, you could start the StalledReview procedure: https://fedoraproject.org/wiki/Policy_for_stalled_package_reviews
As an outsider to Fedora, I don’t know what the problem is, but it seems it’s too much effort to be an “official” Fedora packager. compton has been orphaned for several releases, people are making specfiles and volunteering to maintain the package with no reply. Meanwhile there are three copr repos with the package in it…
Re: compton package
There is https://bugzilla.redhat.com/1051688 with no activity beyond uploading a single package and not keeping it up-to-date. Nobody has contributed any review, not even a self-review or by running the fedora-review tool. No team-building. Where are the other volunteers you speak of?
My point exactly. Maybe this volunteer would be a bit more motivated if someone replied to his review request in the 20 months since he went to the effort of making it?
Packager sponsors like package reviewers have no “first come, first served” queue to work with [1], because it is not a packager sponsor’s obligation to spend time on arbitrary review requests. Packager sponsors choose requests that raise their interest or are within their fields of knowledge. If a ticket is unanswered for months, that is a sign of lack of interest from all sides including users, fellow packagers as well as the submitter, who doesn’t consider it helpful to engage in doing reviews or signing up as a co-maintainer.
[1] http://fedoraproject.org/PackageReviewStatus/
Yes the number of packages in copr goes up, but no all people want pack officially, this my case. I don’t like none crap infrastructure. I do not want be a “official” Fedora packager. COPR, maybe be a final instance for the dead rpmfusion; and all package out-day and the jump to stupid process, for maintain a package or publish new package.
What do you find is bad about the Fedora packaging infrastructure? How do you think it could be better? If you don’t voice your concerns then nobody can address them 🙂
What do you mean with “dead rpmfusion”? They’ve only spent time on setting up modern infrastructure such as Koji instead of Plague. And they package stuff that cannot be included in Fedora and not in Fedora Copr repos either.