TLDR: Need the ability to modify default partitions without having to manually recreate boot. Alternatively need a simple button “create required partitions” (boot) in the manual configuration.
//rant:
The new install wizard, whatever it may be, is quite useless - no way to configure half the options that were available in the old one, and are VITAL to installing an OS.
For example, having to reconfigure hostname post-install is quite silly. Why not have it set up here?
More importantly though, the lack of ability to properly configure the storage is just outrageous. The ability to modify automatically generated partitions or mount points is a must have. You may want /home, but many people don’t. Your use case is probably different from theirs. Give people the freedom back, without having to deal with boot partitions. For instance I for one use VM’s for work a lot and doing this for the different Fedora spins or other things I need to install, is quite silly. I don’t need btrfs on a 20GB virtual drive with split /home do I? Zero benefits, only downsides.
Another good point is that as a community leader, this pains me greatly - not being able to simply guide new users through configuring their storage. This is just rigid and useless. One should not be forced to deal with boot when they want to simply change their /home or anything else that doesn’t touch boot.
Anyway, can we have the old one back, that actually had all the good functionality?
It is never easy to implement a new software. However if you miss the old anaconda, I guess you just have to use the everything ISO. There you still should have the old installer available.
//rant:
I used the the new installer to create a GPT Partition table. The Disk I used was a external USB Disk. It simply deactivated the internal disk in the efi boot options and booting was just possible with the external disk till I activated the disk again in the bios. Might be handy for new users not to overwrite their internal disk. However I needed a moment to find this out.
//rantover
I added the anaconda tag to see if we get seen I could not find it as a title tag.
p.s.
You might check the tag to see that there are already some topics with useful proposals for improvement.
The Fedora Everything iso has the option to change the partitions pretty much like the older version.It is also in the Fedora 44 everything iso. I used it to install Fedora 44 in boxes.
Just today we discussed that we probably should go to the next release party with the talk like “Last call for missing features in the Installer WebUI”, because we currently do not have enough input. Too many people still use the GTK as a fallback option without registering why they needed the fallback first.
So whenever you find that you switch from Web UI to GTK UI for a specific functionality, please report it.
Now, couple of comments:
You can change the hostname in the review screen.
It is not very obvious at this point, we know about this issue. The conversation how to fix it better is in progress, but it is there.
Fedora bugzilla, anaconda-webui component for more straightforward feature requests and bug reports
For more elaborate discussion of the uses cases, with pictures, demos and so on - this forum works well, just use the anaconda tag. I am a PO of the Anaconda team, and I will be using these inputs to for our backlog.
Eventually it will land in Jira but you do not have to interact with it.
IMHO. I we need to follow a video to understand the new “flow”, then is not intuitive enough.
I do agree that there are things that can be done better, but please don’t take the “looking to another side” approach. There have been a decent amount of backlash in reddit, YouTube and here on Discourse to say “there is no enough input”, being the partitioning the main issue.
I think it’s better to hear people and not just wait for bugzilla reports or tickets to improve things, because non-contributors users are not going to open a BZ, they are going to look or something easier to install. Please remember than Linus himself never used debian because the installer was hard to use.
Fedora user input is not the only source of the information we use. But it is an important source, that is why I am worried that we are not getting enough of it. Hence the requests.
Please add specific references to the issues you see.
I do follow some of the threads and social media and discussions. So far all the examples I can remember can be tracked to the Storage Editor being non discoverable at first. Yes, it is a bad thing, but it is not a lack of functionality, it is a lack of discoverability.
For our current work we need to deal with two type of issues:
A) use cases which are functionally impossible, to the extend that you need to use a different tool to perform the installation
B) use cases which are functionally possible, but UI/UX-wise are hard, or confusing.
Non-discoverable hostname and storage editor buttons are the issues of the type B. They are valid issues, but with a workaround. We plan to work on them.
But I am calling for more issues of the type A, because this is the where the hidden risk is the highest. We are worried that we do not know what we do not know. And we can not plan the work for the unknown.
We do need something better than a thread on Distrowatch saying how awful we are. Because we need to understand the use case, ask follow-up questions, figure out how to reproduce and so on.
I am not offering to look at the demo as a defense.
I am asking to understand the problem - was the issue that this path is not working? or was that path not available? or does this path not solve the task?
I haven’t seen the “mount point assignment” option in the ISO I was using - f43-kde (in a VM in this case). From the video I understand that it’s more of a “modify existing drive” situation only. That’s probably why it won’t show up with an empty disk without an existing partition table.
The “old storage editor” had a very easy flow (but perhaps not the best interface) of selecting the desired “usage?” of the drive in a dropdown, something along the lines of “btrfs or lvm or standard” where you would then hit the “create automatically” and you would end up with automatically created partitions you can delete, resize, etc. So whether you chose btrfs or lvm or standard, you had the option to then modify it further to your liking.
Functionally I definitely believe that the old one was a lot better, while the blue underlined hyperlink with a dropdown on the left side, wasn’t the best UI…
Anyway, the use cases of type A (functionally impossible) seem to be, from my limited experience:
Empty disk without an existing partition table: Install a btrfs system with non-standard mount points (e.g without /home or with an extra mount point becaaaause you really want to snapshot only the /usr/lib/libvirt/images or something.)
Empty disk without an existing partition table: Install a new system on an LVM raid array. To complicate it further, with non-standard mount points.
Empty disk without an existing partition table: Install a new ext4 system with encrypted /home only (not / - useful on low power devices running off of SD card but still securing your ssh keys and other personal data.)
Thanks, I am going to double check the hostname part via Jira
I am pretty sure that editable hostname field is implemented in the code, but I need to check what are the conditions for it to show up and why in the case it is not visible.
So it seems to me, all of this configurations should be doable now through the storage editor (the one which is hidden in the menu in the right corner in the WebUI). With a bit more manual steps but possible.
The feature we have in a queue (no assigned target release yet) is to add options to automatic partitioning step so that you can choose between several standard partitioning schemas (BTRFS, LVM, LVM Thin Provisioning..) rather than choose one default which is currently BTFRS for iIve images. It seems more of a design task, because backend functionality for these partitioning schemas exists already.
But as far as I understood, you ask also not just to allow to choose a different default schema, but then also to allow to edit it afterwards?
I think that is a major issue at the moment, the misleading UX aspect: I think I already elaborated it in my last bug report, but the position and the look of this (and the fact that it is there from the first page of the installation) makes users intuitively assume this is something to minimize/maximize the window or maybe to do some other settings of the window, but not that it is related to partitioning.
E.g., when trying to find the settings of a web page, I would never think of checking out the three min/max/exit buttons on the top right, or the “settings” button of firefox itself. Just an example to illustrate what I mean
So some functions people complain about are already there, but we “intuitively don’t find it”. The result is, for many of us (if not most), this function is effectively not available.
Does it maybe make sense to make more tests about UX with a different approach? Not just ask users to click through the installation (which will users not make search for something like that), but tell users which features are there and ask them if they can test these features for you and if the features feel intuitive: so not provide information how to use or find these features, but just mention these features are existing and please test them for you.
E.g., as tests about this occur at the moment: if users now test, they might not find the partitioner, and just assume “ok, that’s not implemented yet I assume”, and then provide +1 feedback, not identifying the issue: they assume its on the roadmap, you assume it’s done → result is that nothing might happen. Maybe there are already other features with that issue too. Just a thought to improve
So it seems to me, all of this configurations should be doable now through the storage editor (the one which is hidden in the menu in the right corner in the WebUI). With a bit more manual steps but possible.
It is not possible without having to manually configure boot. I am talking about the storage editor in the hamburger menu - these options are NOT possible to achieve without having to set up your own boot partitions manually by hand. That’s not acceptable “new user” experience. That is my entire point. Nobody should ever be required to touch that stuff.
You could see it as a feature request for “create boot and let me do the rest”
You are right. And I started the loop of the UI/UX discussion again on this topic after our last interaction, but it is not moving fast yet.
One of the reasons for the delay is that we have several other items in the queue which may or may not change that right corner menu requirements and expectations.
There is a Networking configuration screen, which we tried as an option there, but then decided to move back into one of the wizard steps, there is a Terminal tool, Log Reading tool, and some others. And I expect when we will add them, we will need to redesign that “right corner” concept again.
So while I understand the current frustration, changing it now without a full understanding only to change again in the next release would probably be worse in terms of how much explanation it would need ti the end-user, than to sit on the current implementation just a little bit longer and change it once.
When doing the standard install (at least on kde plasma) once the install is done
I’m told to reboot. But KDE plasma does not have a reboot function.
It’s called “Restart”. Please use consistent jargon.
But what has been asked for before is a button that will do the “Restart” for you,
without the need to hunt through the Live USB’s menus.
True. But thanks for keeping us updated / providing information That’s appreciated.
I wonder if it is useful to create for the time being a type of “common anaconda issues” page, as we have it already for the general for Fedora.
→ This feature is known to be missing and to be implemented
→ This feature is already there at this place, we know already of the UX issue and will adjust the position/location once we have time
→ … …
Stuff like that.
That might help users to avoid some frustration, and avoid your team to get the same reports regularly again. Not sure how easy it is to just refer to it at the beginning of the UX with a link (it’s just text in a window that already contains text?), as it could stand there with a link in a static way until major issues are solved, and then remove the reference. Just an example.