@feborges was talking about this the other day. Let's get him in the loop.

To be clear: I don't know of any pressure from a UX perspective to remove Boxes from the default install, and it's not something I'm particularly pushing for.

I'm -1 on this. I really think that Boxes should be part of the Workstation if we are targeting developers.

Also one question came to my mind while reading this proposal. If we won't be preinstalling Boxes. How we will make sure that the users will find the right application for virtualization (I don't think that they will use Software to find it, then will first open the browser)? My worry is that novice users will just look on Google for some virtualization application and end up with Virtual Box instead.

I agree with @tpopela. It's not possible to support Virual Box as a release blocking VM either, it's not going to get nearly as much testing. If that's the way folks go for virtualization, they're really on their own.

If we won't be preinstalling Boxes. How we will make sure that the users will find the right application for virtualization (I don't think that they will use Software to find it, then will first open the browser)?

Indeed: we ought to consider the app discovery story as part of this. If I do a Google search for "virtual machine fedora", I get:

• Fedora docs on using virt, which aren't helpful at all for a Workstation use case
• OS Boxes - a 3rd party site for downloading OS images
• Fedora Magazine article on virt manager, from last year

The moral of the story - if we want users to use a piece of software, we actually have to document and promote it. Also, it helps if we don't promote the things we don't want to recommend to users.

If I search for "virtual machine" in Software, I get:

• VirtualBox (3 stars)
• Virtual Machine Manager (4 stars)
• Boxes (3 stars)
• Remote Viewer (4 stars)

Lesson: we need to improve app weighting and ratings in Software (that is being discussed currently).

We can at least add banners to Software for the apps we care about, which does give them a bit more exposure. Of course, people need to use Software in the first place for that to work, and it's unclear to what extent people engage with those banners.

All of this is fixable, but not great at the moment.

I agree with @tpopela and am -1. I also agree with @aday that we need to do a much better job of promoting the developer experience we are trying to provide.

I think we have consensus to keep Boxes installed by default. Closing.

Upstream: https://gitlab.gnome.org/GNOME/gnome-build-meta/-/commit/d755534cfbc5f5ec65568bb6fb2e01639ad106a4

Just for the record, late but I am +1 for removing virt from the Workstation Live image.

Reopening. For GNOME 42, we moved Boxes from "meta-gnome-core-utilities" to "meta-gnome-developer-tools." What this means is that while Boxes is still part of GNOME core and built by GNOME release team, it's no longer recommended that we install it by default. Recommend removing from Fedora Workstation default apps to match upstream.

Upstream: https://gitlab.gnome.org/GNOME/gnome-build-meta/-/commit/d755534cfbc5f5ec65568bb6fb2e01639ad106a4

This revert was itself reverted via https://gitlab.gnome.org/GNOME/gnome-build-meta/-/issues/420.

I'm still -1 on removal. We don't have to follow upstream on their organization of applications. In fact, we already don't.

I do not consider gnome-build-meta enough of a reason to remove it from Fedora Workstation. We are not GNOME.

In fact, I'm even considering adding GNOME Boxes for the KDE Spin (fedora-kde/SIG#207) because it's good and app promotion is utter garbage in software centers.

I agree with Neal here. It makes sense not being in gnome-build-meta as a default-install recommendation because GNOME is designed to be generic enough for all vendors/distros and their use-cases.

Fedora Workstation, targeting developers, should override some of those recommendations.

Boxes is an useful tool for third-party developers as well as people developing Fedora itself (since you can express-install Fedora VMs, drag&drop RPM files into it, etc...) which is a common workflow to a lot of packagers I have met over the years.

Fedora Workstation, targeting developers, should override some of those recommendations.

I don't think we want the default installed apps to target developers, though. If so, we'd rename to "Fedora Developer Workstation" and add stuff like Builder. I'd rather provide a leaner starting point while prominently promoting Boxes as the best way to virtualize on Linux.

Fedora Workstation, targeting developers, should override some of those recommendations.

I don't think we want the default installed apps to target developers, though. If so, we'd rename to "Fedora Developer Workstation" and add stuff like Builder. I'd rather provide a leaner starting point while prominently promoting Boxes as the best way to virtualize on Linux.

I think we want to include enough broadly useful things that do support those use-cases, and Boxes definitely fits in that category. A cross section of users we're targeting would all benefit for Boxes: gamers/streamers, and developers.

Conceptually, I have always found some of these fedora-workstation tickets quite depressing and de-motivating.

I think that before discussing additions and removals of applications in public, there should be a private discussion with the upstream maintainer(s) and designers. The public discussion should only happen once there's private consensus.

Remember that many, if not all, of these applications are developed by volunteers in their free time, even if those same volunteers might be paid by a company to work on other things in Fedora. This means that resources are scarce. Sometimes it might take a while for things to take mature, or that things can only happen if we all pull in the same direction.

So, these unexpected public reports, with a lot of back and forth over a low-bandwidth electronic textual medium, feels aggressive and toxic, if you are on the receiving end of it.

I think our biggest fundamental problem is that we use gnome-build-meta as a reason to start these discussions. It's a terrible starting point and really doesn't help our discussions at all.

Fedora is not GNOME, and it's important to remember that. Moreover, our job is to present our best face, and that means offering a high quality selection of tools that make people like our experience and consider it over alternatives. I consider GNOME Boxes to be one of the few unequivocally excellent tools from GNOME, and @feborges is an excellent credit to that.

I agree that Boxes is pretty great. The question is whether it's something that's generally useful to typical non-developer Workstation users, or to developers who do not develop Linux operating systems. I don't think it is. Boxes is a great tool for power users who need it, but I suspect most Fedora users never start it.

We're OK with having small utilities installed by default even if we do not expect them to be used by most users (e.g. Disks, Simple Scan, etc.). Boxes is not a small utility, though.

If we do remove it, then we should promote it more heavily than we do today. If users wind up trying to user Virtualbox instead, they'll be in for a much worse and more difficult user experience.

I think our biggest fundamental problem is that we use gnome-build-meta as a reason to start these discussions. It's a terrible starting point and really doesn't help our discussions at all.

I'll continue to prefer to diverge as little as possible from GNOME. A lot of discussion and consideration goes into deciding which apps belong where. Fedora has become very popular because it sticks very close to upstream defaults.

I agree that Boxes is pretty great. The question is whether it's something that's generally useful to typical non-developer Workstation users, or to developers who do not develop Linux operating systems. I don't think it is. Boxes is a great tool for power users who need it, but I suspect most Fedora users never start it.

We're OK with having small utilities installed by default even if we do not expect them to be used by most users (e.g. Disks, Simple Scan, etc.). Boxes is not a small utility, though.

If we do remove it, then we should promote it more heavily than we do today. If users wind up trying to user Virtualbox instead, they'll be in for a much worse and more difficult user experience.

I think you mean "do not remove it", but yes, I agree it deserves more promotion. I think with our new PRD in mind, we can think about ways to promote Fedora's built-in virtualization capabilities using Boxes.

I think our biggest fundamental problem is that we use gnome-build-meta as a reason to start these discussions. It's a terrible starting point and really doesn't help our discussions at all.

I'll continue to prefer to diverge as little as possible from GNOME. A lot of discussion and consideration goes into deciding which apps belong where. Fedora has become very popular because it sticks very close to upstream defaults.

Most people don't consider app selection one of those things, but I can see the argument there.

I think you mean "do not remove it", but yes, I agree it deserves more promotion. I think with our new PRD in mind, we can think about ways to promote Fedora's built-in virtualization capabilities using Boxes.

Nah, that wasn't a typo. As long as it's there by default, most users who need to virtualize will give it a shot, no need to mess with Virtualbox unless they have trouble with Boxes. But if it's no longer installed by default, this may no longer be true unless we promote Boxes heavily (e.g. big banner in Software). We still want users to find Boxes because it is part of our developer workstation strategy, just like Builder, and is way easier and more convenient to use than Virtualbox.

Agreed: Workstation WG still wants to keep Boxes.