#172 Reallocation and removal of applications.
Closed: Won't fix 3 years ago by aday. Opened 3 years ago by johannbg.

I'm noticing a lot of removal of "default" application as in issues #164 #166 #169 #171

Back in the day when I wore my Fedora ambassador hat and was involved with installing and setting up Fedora on regular end users computers one of the most complaint issue was the re-allocation of applications which required the end users to "re-train" where they should go to find that application they regularly used or direct removal of the application they routinely used ( during upgrades ) which caused end users to stop using not only Fedora but Linux altogether.

Now Gnome developers (upstream or here downstream ) and or designer may not be aware of or realize that it's exactly these types of unstable UI design and disruptive changes like removal or re-allocation of application that lead to regular end users choosing not use Fedora, Gnome or Linux et all something that should have been very much apparent when Gnome transitioned from 2.x to 3.x. and all the screams of Gnome users pain echoed through the internet.

To shield at least Fedora users from this painful experience and indecisiveness of upstream/downstream which applications should be installed by default or exist and where they should be located I propose that Fedora's Gnome simply be stripped of all "default" application which eliminates this problem completely and end users simply be presented with additional step in Gnome initial setup in which the end user selects the application he or she wants to use from the market place which will then be tailored to his or her foreseeable workload and computer usage ( which will also familiarize the end user immediately with the market place since the end user has no other option but to install applications he or she uses ) instead of either upstream/downstream trying to decide that for the end users which it never gets right or upstream constantly removing or re-allocating application because it cant decide where the application should be placed or which application should be installed and or part of the "default" experience.

Given that neither we nor upstream are getting better ( and are at conflicts what the default end users experience should be ) at this after decade(s) it should be quite apparent that we need to change how we approach things as in stop deciding what we think the end user wants and instead simply allow him or her chose what application he or she actually wants and will use and provide him or her simply with "recommendations" in the market place itself.


Note that the changes listed above aren't necessarily all going to happen at once, and some are just things that we're considering.

I think that we all agree that we want to avoid disruptive changes for our users. However, I'm not sure that concern applies in this case, for a few reasons.

First, this is removing or changing the apps that are included in the default install, so it's not going to affect users who upgrade from one version to the next - the UI isn't going to change out from under someone.

Second, if someone finds that an app is missing that they want, they can install it - it's not a major disruption. We actually have a fairly good path for that through system search.

Third, the apps that we're looking at are generally minor and not heavily used - Cheese and Weather, for example. Others (Photos and Rhythmbox) don't get used as much as they used to.

I think we definitely agree that we don't want indecisiveness around the default app set, and that's one reason why we're tracking these changes and working on guidelines for what should be included and why (that's #125).

I see various issues with making app installation completely user-driven (as opposed to having a default app set). A main one is that users would end up having to install a lot of apps on the fly, as they found they needed them (which can particularly be an issue in offline scenarios), rather than having a more complete out of the box solution. Another issue is that we'd lose the ability to provide a default experience that is both integrated and consistent.

I see various issues with making app installation completely user-driven (as opposed to having a default app set). A main one is that users would end up having to install a lot of apps on the fly, as they found they needed them (which can particularly be an issue in offline scenarios), rather than having a more complete out of the box solution. Another issue is that we'd lose the ability to provide a default experience that is both integrated and consistent.

Well the whole idea would to be stop trying to provide a default experience that in reality has never been integrated and consistent over the years due to the fact that applications have been re-allocated and or removed right and if the focus is supposed to be on flatpak and accommodating marketing place that also means
a) there is already an expectation that there is some form of network connectivity in place
b) that Gnome's own upstream applications should be under the same "marketing" laws as any other application available at the market place ( thus subjected to install/removal at end users behest ) since end users might want to install something more "popular" or with a higher reviews and ratings and even pay for an app with better support than anything that Gnome community might be writing/maintaining.

If we take "Rhythmbox" as an example a decade ago it might have had it's part to play, today end users are more likely expect or to install something like Spotify which is available on all their devices ( mobile phone/tv/car audio devices/home audio devices ) as opposed an "integrated" music playing application that came with the desktop environment.

The cold hart reality is that we for the past decades have spent tremendous energy upstream/downstream trying to define what a "default" experience should be for end users, tried to define what that typical end users looks like and "guessing" what he or she might have installed and running in endless circle chasing our tails to that end by adding or removing application or application stacks meanwhile the world has evolved to a completely different set of end users expectation which today they expect having a user account in associated market place which keeps tabs on which application they use on which devices and easily allows them to install it and remove across their devices, which in turns provides them with that integrated and consistent desktop experience and feel across their devices.

So arguably we should stop spending all that energy upstream/downstream chasing that end users end of the rainbow which involves providing that default end user experience from out of the box and simply focus more on the end user experience in the application market place so the end user can determine that for him or herself and gives him or her that integrated and consistent desktop experience and feel.

Today end users focus more on the availability of applications rather than which platform they are currently running on ( OS-X,Windows,Linux or Linux OS variant Arch/Debian/Fedora/Suse etc ) so arguably if Fedora/Gnome wants to stay relevant it needs to stop focusing on being "traditional" and be more "modern" and focusing more on the core desktop environment stability and performance and the market place experience ( no more rpms, only flatpak ) while allowing the end user determine which applications suit him best thus solving that never ending discussions and dispute of what should or should not be installed by "default".

I definitely agree that we ought to be focusing on the app marketplace as you described it, and the issues you mentioned in the report are about slimming down the default app set because of some of the factors you've described.

However, we still need default apps for some things. You still want to have a browser, handlers for the basic file types, basic utilities, and so on. Better to have those things working out of the box than have to hunt around for them.

One should be able to install Fedora without internet access and be able to perform basic tasks such as visualizing an image, playing a video, etc....

Well I would disagree and the browser we are shipping is the wrong one from end users perspective since Chrome has clear marketing dominance [1] so it's higher probability that end users are using Chrome as their preferred choice since it integrates better with the entire Google ecosystem ( Android,youtube,gmail etc ) and we do them disservice by installing Firefox by default which forces them to install Chrome instead.

Then there is the whole legal issues surrounding browser which could applicable to other applications as well. I'm not sure why Linux distribution somehow manage to escape the "Browser Choice screen" that EU ruled that Microsoft had to offer end users back in the day.

The majority of people here use online personal information management application that provides integrated mail, calendaring and address book functionality like Gmail or Microsoft 365 365 ( or some locally run groupware solution ) which is probably the same across the world so applications like Evolution and Thunderbird are becoming obsolete.

People should not have to "hunt" for application if handlers for the basic file types should just trigger "open in app" type of behaviour and offer to install it which is what Google does on the Android platform ( if you try to play a youtube video in the browser it first tries to open that in the "youtube" app and offers to install it if it's not available instead of immediately start playing it ) and what are "basic utilities" and belong those "basic utilities" in a separated app as opposed to be tightly integrated with Gnome?

Probably most of the "basic utilities" belong in a "File Manager" which is interchangeable on most platforms as are disk tools, calculators, clocks and weather applications as well so what is considered "basic utilities" and why do those have to be installed by default instead of "on demand".

  1. https://www.w3schools.com/browsers/

One should be able to install Fedora without internet access and be able to perform basic tasks such as visualizing an image, playing a video, etc....

Well you can install Fedora without internet access and the installer will need to be able to install flatpaks instead of rpms in the future but why would you install/setup a computer on the 21 century that has no network connectivity?

We are long past the days of having to install a linux distribution off multiple floppy disks and living in the middle ages so what's the usecase here?

End users dont run around with portable disk drives and hook them up to the computer to play music and videos when they live in an era when a year subscription with netflix and spotify is probably cheaper than the hardware required to do so and they can play it from any device they own.

  • That already contradicts the workstation default experience since it offers users to use the single sign-on framework for GNOME and enables bunch of type services which require network connectivity to work out of the box then there are things like updating the firmware and software on their computer.

I agree with Allan and Felipe. Switching to a minimal install without any apps would not be a very welcoming user experience. And it's easy to uninstall apps you don't want.

Of course, we should continue to continually reconsider whether the apps we are installing are truly useful, to ensure we're not shipping bloat.

Thanks for the feedback; let's close this for now.

Metadata Update from @aday:
- Issue close_status updated to: Won't fix
- Issue status updated to: Closed (was: Open)

3 years ago

Just bear in mind this constants fiddling with which applications are installed either needs to stop or be solved differently since we are wasting tremendous energy in guessing what end users want or need and are doing end users a great disservice by constantly removing,replacing, re-allocating applications in the DE so in the future let's try working towards a solution that stabilizes this for the end users as well as minimizes the workload on everyone involved every cycle.

It's been a little while since we've made changes to the default apps. I think the last change was removing Documents. Calling it "constant fiddling" seems a little extreme. We make changes here or there, but overall the set of default apps has been very consistent for the past five years. It's just been a while since we last did a thorough overview and reconsideration of what we're installing.

Perception is a funny thing.

That said there is a high probability that Gnome evolves in the same direction as AOSP ( depends on how successful the application market will become ) which would mean that Gnome and the other DE's will eventually stop being involved in the development of the usable set of applications that would be shipped but instead just provide functionality to other applications in the application market that full fill that role and serve as a demonstration for the app developer community of the market and the DE.

So arguably the goal for Fedora should simply be that the desktop environment will only ship application that are only installable and removable by the app market ( if an application does not meet that requirement it should not be part of the default experience or available in the distribution as an rpm package either but that will probably cause a huge uproar ) and then just ship whatever set of application upstream is shipping by default so we evolve with upstream otherwise we just end up competing against upstream which is a battle we can never win.

There was a tweet the other day [1] in which an individual had booted GNOME OS image from git on native hardware and he talked about how super snappy it was since it did not contain the "normal distro junk" which is a clear indicator that downstream shipping Gnome are negatively affecting the overall desktop experience of Gnome which arguably is the opposite effect of what downstream should be doing ( it should match and or improve the upstream experience ) and upstream would want since it not only reflects badly on the downstream distribution ( since it cant match the upstream experience ) but also on upstream ( since the desktop experience of Gnome should be the same or better regardless of the distro shipping it ).

  1. https://twitter.com/hergertme/status/1286014838179655685

There was a tweet the other day [1] in which an individual had booted GNOME OS image from git on native hardware and he talked about how super snappy it was since it did not contain the "normal distro junk"

I've been mostly successful in keeping upstream and downstream default apps in sync for the past several years. The GNOME OS images currently contain one or two apps that aren't supposed to be there, for which I've already reported bugs. Other than that, they should be almost the same as Fedora. We swap Epiphany for Firefox, swap Music for Rhythmbox, and add LIbreOffice and ABRT. That's it. I think Firefox and LibreOffice are here to stay. ABRT is under discussion. We've tentatively approved removing Music, but that still needs upstream discussion as well.

So we're basically already sticking with upstream defaults wherever possible. LibreOffice is too important to get rid of, even I'll admit Epiphany is not yet ready, and ABRT is a bit special. Doesn't make sense to categorize any of this as bloat.

Login to comment on this ticket.

Metadata