So this is probably just the lack of sleep talking, but I notice a pattern in the comments about newUI is that people really aren't confident about which partitions exactly form part of the install, in newUI custom part
I was thinking, could we emphasize the difference between 'new Fedora 18 install' and 'existing detected installs' in the left-hand pane more heavily? the specific idea I had may be silly, but it's color coding. give the 'new Fedora 18 install' tree a green background and the old detected installs grey backgrounds, or something. would that help make things clearer?
So this is probably just the lack of sleep talking, but I notice a pattern in the comments about newUI is that people really aren't confident about which partitions exactly form part of the install, in newUI custom part
Can you provide a little more detail here? Is there confusion over the very concept, or over what's new and what's old, etc.?
I was thinking, could we emphasize the difference between 'new Fedora 18 install' and 'existing detected installs' in the left-hand pane more heavily? the specific idea I had may be silly, but it's color coding. give the 'new Fedora 18 install' tree a green background and the old detected installs grey backgrounds, or something. would that help make things clearer?
Well, the indentation is supposed to help group things together. I think we could also color code things, perhaps. That's always tricker than I want it to be. It could either be new vs. old, or every other expander could be shaded like you might seen in a spreadsheet. We could perhaps also put a horizontal bar between the new one and the old ones.
- Chris
On Wed, 2012-11-14 at 18:15 -0500, Chris Lumens wrote:
So this is probably just the lack of sleep talking, but I notice a pattern in the comments about newUI is that people really aren't confident about which partitions exactly form part of the install, in newUI custom part
Can you provide a little more detail here? Is there confusion over the very concept, or over what's new and what's old, etc.?
What I've seen on the forums is that people just don't quite get the concept of the custom part screen - that you have partitions down the left, including existing partitions and newly created ones, and that the workflow is to create a partition and then define its properties and assign it to a disk and so on.
On Nov 23, 2012, at 7:26 PM, Adam Williamson awilliam@redhat.com wrote:
On Wed, 2012-11-14 at 18:15 -0500, Chris Lumens wrote:
So this is probably just the lack of sleep talking, but I notice a pattern in the comments about newUI is that people really aren't confident about which partitions exactly form part of the install, in newUI custom part
Can you provide a little more detail here? Is there confusion over the very concept, or over what's new and what's old, etc.?
What I've seen on the forums is that people just don't quite get the concept of the custom part screen - that you have partitions down the left, including existing partitions and newly created ones, and that the workflow is to create a partition and then define its properties and assign it to a disk and so on.
Oldui partitioning visually took up more than 1/2 of the screen, and this same information is compressed into 10% of my screen while not conveying the resulting partitioning.
Oldui's modal dialog containing the details of each partition (file system, size, label) was 10% of the screen real estate, and is now greater than 50% and no longer modal.
The result is a shift of available space, and hence emphasis, on details rather than big picture. And I think it's increasingly clear that most users need more big picture, than a huge amount of estate reserved for details. Also, newui is becoming increasingly wordy, with excessive amounts of text to explain things, which again emphasizes detail over big picture.
Chris Murphy
On Fri, 2012-11-23 at 20:57 -0700, Chris Murphy wrote:
On Nov 23, 2012, at 7:26 PM, Adam Williamson awilliam@redhat.com wrote:
On Wed, 2012-11-14 at 18:15 -0500, Chris Lumens wrote:
So this is probably just the lack of sleep talking, but I notice a pattern in the comments about newUI is that people really aren't confident about which partitions exactly form part of the install, in newUI custom part
Can you provide a little more detail here? Is there confusion over the very concept, or over what's new and what's old, etc.?
What I've seen on the forums is that people just don't quite get the concept of the custom part screen - that you have partitions down the left, including existing partitions and newly created ones, and that the workflow is to create a partition and then define its properties and assign it to a disk and so on.
Oldui partitioning visually took up more than 1/2 of the screen, and this same information is compressed into 10% of my screen while not conveying the resulting partitioning.
Oldui's modal dialog containing the details of each partition (file system, size, label) was 10% of the screen real estate, and is now greater than 50% and no longer modal.
The result is a shift of available space, and hence emphasis, on details rather than big picture. And I think it's increasingly clear that most users need more big picture, than a huge amount of estate reserved for details. Also, newui is becoming increasingly wordy, with excessive amounts of text to explain things, which again emphasizes detail over big picture.
Funnily I was kind of thinking the opposite - we don't have _enough_ text. I'm no design expert so I don't know if it's considered 'old fashioned' or 'undiscoverable' to have a simple line or two explaining how a screen actually works - maybe it turns out people don't read such stuff, I don't know - but to me, newUI seems to be missing some such things. The custom part screen has no explanation at all, it just sits there and you try to figure it out. Another little thing which confused at least one user: when creating a partition, what do you do in the size field? Technically it's awesome - it understands lots of different ways of describing a size. But it doesn't _tell_ you that, so you don't _know_ what to type, you're worried that whatever you type will be wrong or will break it. I know I found it a bit confusing at first. (Note I haven't really look at post-Beta stuff yet, so all of this post relates to what's in Beta; if any of this stuff has been improved post-Beta, I'm not aware of that yet).
Here's an interesting forum thread:
http://forums.fedoraforum.org/showthread.php?t=286162
ignoring the bit where I blow my top somewhat unfairly at Dan, note the utterly incorrect interpretation of the custom partitioning screen that the OP managed to come up with. It's apparent that at least one person can look at the custom part screen and imagine that it works a completely different way from how it _actually_ works, with the obvious attendant frustration.
For those who don't have time to clicky, here's an excerpt:
"I'm guessing here: This version of Anaconda seems to read the labels of existing partitions off the hard drive and, again I'm guessing here, maybe in conjunction with which existing Linux installations have GRUB installed, presents you with scenarios that it thinks would work, i.e. partitions with enough room to install Fedora.
In the end, instead of choosing a disk partition by using standard "sdxy" nomenclature, you have to know the label of the partition you want to use (in my case, sda11 was labelled "test2") and tell Anaconda to use that as "/". Choose another partition label to use as "/home" if you wish. Then you get to tell Anaconda where to mount the other partitions, all by their labels."
This is obviously not correct. :)
One detail from that thread which is becoming a constant refrain on the forums is that people are confused when they go to create a partition and all they get is 'mount point' and 'size'. Apparently they don't grok the 'reverse workflow' and think 'obviously I'll get to decide what type of partition it should be and where it should go later', it seems a lot of people instead just think 'oh my god! I can't decide what type of partition it should be or where it should go!' I think the initial simple dialog implants that thought in their head, and then they just don't think to look around and find the + expander that exposes most of those options.
Hope this doesn't come across as overly negative, and of course there is positive feedback on various elements of newUI too - I'm just isolating some of the negative feedback to identify areas we can improve on.
On Tue, 2012-11-27 at 01:39 -0800, Adam Williamson wrote:
On Fri, 2012-11-23 at 20:57 -0700, Chris Murphy wrote:
On Nov 23, 2012, at 7:26 PM, Adam Williamson awilliam@redhat.com wrote:
On Wed, 2012-11-14 at 18:15 -0500, Chris Lumens wrote:
So this is probably just the lack of sleep talking, but I notice a pattern in the comments about newUI is that people really aren't confident about which partitions exactly form part of the install, in newUI custom part
Can you provide a little more detail here? Is there confusion over the very concept, or over what's new and what's old, etc.?
What I've seen on the forums is that people just don't quite get the concept of the custom part screen - that you have partitions down the left, including existing partitions and newly created ones, and that the workflow is to create a partition and then define its properties and assign it to a disk and so on.
Oldui partitioning visually took up more than 1/2 of the screen, and this same information is compressed into 10% of my screen while not conveying the resulting partitioning.
Oldui's modal dialog containing the details of each partition (file system, size, label) was 10% of the screen real estate, and is now greater than 50% and no longer modal.
The result is a shift of available space, and hence emphasis, on details rather than big picture. And I think it's increasingly clear that most users need more big picture, than a huge amount of estate reserved for details. Also, newui is becoming increasingly wordy, with excessive amounts of text to explain things, which again emphasizes detail over big picture.
Funnily I was kind of thinking the opposite - we don't have _enough_ text. I'm no design expert so I don't know if it's considered 'old fashioned' or 'undiscoverable' to have a simple line or two explaining how a screen actually works - maybe it turns out people don't read such stuff, I don't know - but to me, newUI seems to be missing some such things.
What about adding a little [?] button at the bottom of every spoke that would explain how spoke works? If some warning needs to be shown (the orange box we have now), it can go next to that button.
Just my 2 cents.
On Tue, Nov 27, 2012 at 01:39:14AM -0800, Adam Williamson wrote:
One detail from that thread which is becoming a constant refrain on the forums is that people are confused when they go to create a partition and all they get is 'mount point' and 'size'. Apparently they don't grok the 'reverse workflow' and think 'obviously I'll get to decide what type of partition it should be and where it should go later', it seems a lot of people instead just think 'oh my god! I can't decide what type of partition it should be or where it should go!' I think the initial simple dialog implants that thought in their head, and then they just don't think to look around and find the + expander that exposes most of those options.
Hope this doesn't come across as overly negative, and of course there is positive feedback on various elements of newUI too - I'm just isolating some of the negative feedback to identify areas we can improve on.
Are these typically new users, or people who have years of experience with oldui? I think it should be more intuitive for new users, and I can see how old users are trying to find something familiar and failing :)
What I don't want to see us doing is starting to patch the ui with hints here and there, or big blocks of text that are never read. Maybe what we need are some blog posts going through example scenarios to help familiarize users with the new way of doing things.
On Tue, 2012-11-27 at 10:13 -0800, Brian C. Lane wrote:
On Tue, Nov 27, 2012 at 01:39:14AM -0800, Adam Williamson wrote:
One detail from that thread which is becoming a constant refrain on the forums is that people are confused when they go to create a partition and all they get is 'mount point' and 'size'. Apparently they don't grok the 'reverse workflow' and think 'obviously I'll get to decide what type of partition it should be and where it should go later', it seems a lot of people instead just think 'oh my god! I can't decide what type of partition it should be or where it should go!' I think the initial simple dialog implants that thought in their head, and then they just don't think to look around and find the + expander that exposes most of those options.
Hope this doesn't come across as overly negative, and of course there is positive feedback on various elements of newUI too - I'm just isolating some of the negative feedback to identify areas we can improve on.
Are these typically new users, or people who have years of experience with oldui? I think it should be more intuitive for new users, and I can see how old users are trying to find something familiar and failing :)
What I don't want to see us doing is starting to patch the ui with hints here and there, or big blocks of text that are never read. Maybe what we need are some blog posts going through example scenarios to help familiarize users with the new way of doing things.
I believe it mostly is people who are used to oldUI, yeah. Still, that's a large chunk of our users, and it's still something to try and resolve.
On Tue, Nov 27, 2012 at 7:40 PM, Adam Williamson awilliam@redhat.com wrote:
I believe it mostly is people who are used to oldUI, yeah. Still, that's a large chunk of our users, and it's still something to try and resolve.
Yeah, that's the main criticism I've also seen toward gnome. You can try attracting new users, but if you lose many of the old ones in the process it can be a negative balance.
-- Gianluca Sforna
http://morefedora.blogspot.com http://identi.ca/giallu - http://twitter.com/giallu
On Nov 27, 2012, at 2:39 AM, Adam Williamson awilliam@redhat.com wrote:
kind of thinking the opposite - we don't have _enough_ text. I'm no design expert so I don't know if it's considered 'old fashioned' or 'undiscoverable' to have a simple line or two explaining how a screen actually works
The instant you need text to explain how a GUI works, your GUI is a failure.
Text can help set context where a GUI can't. But that isn't the problem with newui. People understand the context, they're just getting lost in the GUI.
Chris Murphy
On Tue, 2012-11-27 at 11:29 -0700, Chris Murphy wrote:
On Nov 27, 2012, at 2:39 AM, Adam Williamson awilliam@redhat.com wrote:
kind of thinking the opposite - we don't have _enough_ text. I'm no design expert so I don't know if it's considered 'old fashioned' or 'undiscoverable' to have a simple line or two explaining how a screen actually works
The instant you need text to explain how a GUI works, your GUI is a failure.
Like I said, I am not a design expert and I'm not familiar enough with the state of such arguments to participate. I've noticed this is a definite trend in all modern software design, but I do not know if it's a universal constant which can be stated with such authority, especially as it relates to complex operations.
My layman's understanding would be that you can make a relatively simple operation / set of operations more or less entirely intuitively discoverable, or you can make a complex set of operations intuitive _gradually_, by introducing a single element at a time (my silly example of this would be that Nintendo are great at doing it as a part of game design - you start off learning to jump and suddenly ten hours later you're riding a horse and simultaneously shooting things out of the air with a zoomable bow-and-arrow interface, or something). But I cannot right now think of an example of a GUI which makes a complex set of operations entirely intuitively discoverable when that operation is to be performed in a single go - which is what the custom partitioning screen does. Does software like, say, an office suite (used at an advanced level, not just a 'type this on the screen and print it' level), an advanced image editor, or an advanced video editor follow this dictum? If not, is that a function of bad design or an inevitability given the complexity of the tasks they handle?
Still, I don't want to veer off into the theoreticals too much as it'd be getting off track and I'd be bringing the metaphorical knife to a gun fight. If your dictum is the one the newUI design is being done by, then great: we don't need more text, but we _do_ need to make the design more discoverable somehow, because it demonstrably isn't entirely so right now.
On Tue, Nov 27, 2012 at 10:50:54AM -0800, Adam Williamson wrote:
On Tue, 2012-11-27 at 11:29 -0700, Chris Murphy wrote:
On Nov 27, 2012, at 2:39 AM, Adam Williamson awilliam@redhat.com wrote:
kind of thinking the opposite - we don't have _enough_ text. I'm no design expert so I don't know if it's considered 'old fashioned' or 'undiscoverable' to have a simple line or two explaining how a screen actually works
The instant you need text to explain how a GUI works, your GUI is a failure.
Still, I don't want to veer off into the theoreticals too much as it'd be getting off track and I'd be bringing the metaphorical knife to a gun fight. If your dictum is the one the newUI design is being done by, then great: we don't need more text, but we _do_ need to make the design more discoverable somehow, because it demonstrably isn't entirely so right now.
The design is discoverable, but I think the problem is that after so many years of bottom-up construction they are confused by the new way of thinking about things.
When you do complex operations do you just dive right in? Or do you try to learn how to use it first? It isn't always possible to make complex things intuitive. I think that much of this could be fixed by documentation and glossy photographs with circles and arrows and a paragraph for each one.
On Tue, 2012-11-27 at 11:58 -0800, Brian C. Lane wrote:
On Tue, Nov 27, 2012 at 10:50:54AM -0800, Adam Williamson wrote:
On Tue, 2012-11-27 at 11:29 -0700, Chris Murphy wrote:
On Nov 27, 2012, at 2:39 AM, Adam Williamson awilliam@redhat.com wrote:
kind of thinking the opposite - we don't have _enough_ text. I'm no design expert so I don't know if it's considered 'old fashioned' or 'undiscoverable' to have a simple line or two explaining how a screen actually works
The instant you need text to explain how a GUI works, your GUI is a failure.
Still, I don't want to veer off into the theoreticals too much as it'd be getting off track and I'd be bringing the metaphorical knife to a gun fight. If your dictum is the one the newUI design is being done by, then great: we don't need more text, but we _do_ need to make the design more discoverable somehow, because it demonstrably isn't entirely so right now.
The design is discoverable, but I think the problem is that after so many years of bottom-up construction they are confused by the new way of thinking about things.
When you do complex operations do you just dive right in? Or do you try to learn how to use it first? It isn't always possible to make complex things intuitive. I think that much of this could be fixed by documentation and glossy photographs with circles and arrows and a paragraph for each one.
Right, that is my inclination too and hence my initial mail suggesting MOAR TEXT, but I'm willing to recognize you and I may be the aberrations and other people just buy a chainsaw, throw away the manual, point it at a tree and start poking buttons :) Whether we aim to document the approach and rely on people reading the instructions, try and clarify the approach through UI, or do both, I don't mind.
On Nov 27, 2012, at 12:58 PM, Brian C. Lane bcl@redhat.com wrote:
The design is discoverable, but I think the problem is that after so many years of bottom-up construction they are confused by the new way of thinking about things.
I refuse the premise that it's oldui familiarity resulting in newui confusion. I'm new to both oldui and newui and find oldui immensely more discoverable and navigable than newui with respect to custom partitioning.
At least as big a problem for the present state of testing and opinions of newui is that that anaconda ui/ux itself hasn't been stable. There have been significant changes, not merely bug fixes, pre-alpha F18 to beta final.
When you do complex operations do you just dive right in? Or do you try to learn how to use it first? It isn't always possible to make complex things intuitive.
I dive right in. I regress when I encounter a problem.
I think that much of this could be fixed by documentation and glossy photographs with circles and arrows and a paragraph for each one.
Because you used the word glossy, I'm going to assume this whole paragraph is hyperbole. That this much effort would be necessary to describe UI function would by my definition make the UI a hostile user experience.
Chris Murphy
anaconda-devel@lists.stg.fedoraproject.org
-
Adam Williamson -
Brian C. Lane -
Chris Lumens -
Chris Murphy -
Gianluca Sforna -
Vratislav Podzimek