Jono Bacon is editorializing on his blog regarding usability on FOSS applications. UPDATE: Acts of Volition looks at how “creeping featuritis” has affected interface design in different open source software projects. The article notes that while some projects have succumbed to the problem, others have managed to tame the beast.
Refining for usability; The Rise of Interface Elegance in OSS
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
While what he sais is true, this has been said a hundred times before, still nothing happends.
Hopefully his little editorial will raise a few heads, but I somehow doubt it.
There’s still no DE on linux that is as light and easy to use as BeOS IMO. I seriously wonder why. Because something similar to the tracker/deskbar wouldn’t be that hard to replicate.
>There’s still no DE on linux that is as light and easy to >use as BeOS IMO. I seriously wonder why.
BeOS was lite because it was an integrated product created from scrarch, legacy-free, by the same team. They had clear goals from the beginning and its engineers’ cubicles were next to each other to discuss architecture (as JBQ said in an os news article).
Any F/OSS OS project on the other hand is a lot of cross and stitch job, it _can not_ be made as lightweight and as integrated as a product created by a small team working together every day. Not on the scale that linux or BSD is today anyway with all the legacy they carry over from Unix and other technologies over the years.
This is not to say that Linux doesn’t have other advantages though, because it does. It’s just that these advantages are not in the domain of having a “lightweight” OS. If they are to do that, a lot of application and API compatiiblity would have to be thrown out of the window and that would result on other problems. Linux has other advantages with its open development status, the good networking support, some vendor support etc, things BeOS never had.
So, just use best tool for the job and whatever finds you in agreement with the development model.
>as JBQ said in an os news article
Actually, I wrote that, quoting JBQ.
http://www.osnews.com/story.php?news_id=6574
Even more interesting then the article, IMHO, are the comments on it from people like jdub.
Make sure you read them!
BeOS was lite because it was an integrated product created from scrarch, legacy-free, by the same team.
Well, I was only talking about the DE here, not the whole OS.
The Tracker/Deskbar isn’t really tightly integrated.
While XFCe and IceWM both are light, they are lacking ease of use and elegance.
I’m not saying that the Tracker/Deskbar is superiour, it lacks some neat features, but i find it generally much more usable than any DE available for linux.
The KISS principle seems forgotten, even in the light DEs.
Just wanted to write the same…
From what I remember, those are some of the reasons why user levels (which were tried in GNOME before) aren’t that great of an idea:
– People tend to overrate themselves.
– It makes life very hard in case you need a setting which isn’t available in your user level.
– Not every experienced user enjoys being flooded with settings.
– It certainly makes it harder to design one interface that just works right for everyone (including the right mix of preferences).
There might be more and I’m no usability expert…
The author clearly doesn’t seem to know GNOME that well, because GNOME certainly doesn’t try to please everyone.
The source of many flames… GNOME only tries to please people who want things to “just work”, whether they are novices or professionals who need a job done. It doesn’t try to please those who enjoy tinkering with the desktop itself. And I believe that they are doing a damn fine job. While there are still some rough edges, I have been nothing but pleased with the development since GNOME 2.0.
Also he wrote: “We have a difficult decision to make here. Someone has contributed a patch that they have spent their time and effort creating, but the patch adds a feature that may detract from the general direction that the project is heading in. Do you allow the patch and possibly bloat the project up unnecessarily, or do you say no and possibly alienate a contributor? It is a difficult decision to make, and there is no clear option either way.”
I haven’t seen any serious project lately accepting patches just to please a contributor. Just ask Havoc what he thinks about this statement. 😉 So I don’t think that this is a difficult decision at all.
>The Tracker/Deskbar isn’t really tightly integrated.
Actually it is. Deskbar “understands” Tracker.
User levels are a good idea. if done right.. It failed in Nautilus.. it worked in Apple’s “simplified finder” (when AtEase came out)..
I think the KDE folks should focus in the KIOSK mode. It’s more or less achieves the same purpose as Apple’s AtEase program, that was a HUGE success in certain places, especially K-12 education. We used to use it back on two macs back in 1997 or so in an elderly day care center and it worked with people who were downright uncomfortable with the vanilla MacOS 7.5 (imagine putting them in front of KDE or GNOME!)
Some of the settings in some KDE apps is redicioulous for even the toughest of geeks, like me. Take, in Konqueror, for example, the “Crypto” module in the Settings dialog. It has 6 tabs full of arcane terms.. the first tab has a huge list of Ciphers. I know what a cipher is (most people don’t), but hell if I know what the difference between “EXP-RC2-CBC-MD5” and “EXP-RC4-MD5” is. I’ve used KDE since KDE 2.0, and have seen this dialog gradually expand over time.
Basically, unless you have training in cryptography, you won’t understand 75% of what this module says..
Nothing really new here. The novice, intermediate, advanced user schema has been bantered around before, including here. I don’t think it would take any major architectural changes in KDE to make this happen.
In any case, this is what for distros are for. I would guess that the majority of KDE users are satisfied with menu/feature bloat.
I’m a big fan of plugins. I believe they should be used as much as possible. He’s right that not every patch should be included into a core just because the code is technically sound. Plugins are actually used quite a bit in the KDE architecure. Isn’t KParts really a plugin? Kdevelop makes extensive use of plugins from what I’ve seen. I guess the question is what kind of interesting things can you do if you took plugins to an extreme.
I would recommend checking out Eclipse to see what can be done with a powerful plugin framework.
I would guess that the majority of KDE users are satisfied with menu/feature bloat.
Well, yeah, because the ones that aren’t (like me) has allready switched to something else. I would use KDE if it wasn’t that bloated.
Though, I still use some KDE apps, like Quanta and KWrite, anyone have some alternatives to recommend? And please don’t say Bluefish and Emacs.
Actually it is. Deskbar “understands” Tracker.
What I meant was that they aren’t tied to the rest of the system, and the system isn’t tied to them. You can easily make replacements for them. Though, some apps will require libtracker still.
No, it can be done, but KDE and Gnome are examples of how not to do it.
From what I have seen, they put way to much emphasis on using xml, which generally does not solve any problems but just adds extra weight, more wrapping than what is necesary, creating only X programs.
Linux and all other systems are not meant to run throught X, X is only meant to be a front end, when you have something that can’t be done on a CLI, you have a program that is bad and needs reworking.
“While what he sais is true, this has been said a hundred times before, still nothing happends.
Hopefully his little editorial will raise a few heads, but I somehow doubt it.
There’s still no DE on linux that is as light and easy to use as BeOS IMO. I seriously wonder why. Because something similar to the tracker/deskbar wouldn’t be that hard to replicate.”
What are you talking about, freedesktop.org has been working on the issue of everyone doing everything different (inconsistency). The Gnome HIG is largely for that purpose, and it works quite well if you only use Gnome applications. Even KDE claims to be doing this, and I think KDE apps are decently consistent.
Heads have been raised, but with thousands of programs and many authors who disagree on what is good I don’t think we are gonna see the spitshine polish that people only expect from Apple anyway.
Oh, and Xfce. It’s light, it has gui config for everything except startup. Gnome is very easy to configure, although I doubt anything is as responsive as BeOS (I understand that its KERNEL is renowned for responsiveness).
People are too picky. You could be handed perfection and you would complain its hardware is too expensive.
Hi
“Though, I still use some KDE apps, like Quanta and KWrite, anyone have some alternatives to recommend?”
kate, quanta and k3b has basically no good equivalents.
What are you talking about, freedesktop.org has been working on the issue….
Even though the issue has been raised doesn’t mean that they are approaching it the right way IMO. Now there are many right ways as there are many different people. What I meant was that I still haven’t found a DE for linux that is as simple and easy to use as BeOS is to me, and I know a lot of people that agree with me.
While XFCe might have features for most stuff, the way it provides them are kinda akward. If I want to add a launcher I just want to D&D the app onto the panel and it will give it the right icon and name automagically. But no, I have to add a launcher by rightclicking and hunt down the binary, give it a name and icon myself.
Also the items in the panel are positioned using a positing number that’s available in the item’s config panel. Again, what’s wrong with D&D? Not light enough?
And what’s the deal with “Attach menu to launcher” which is the only way I’ve found to create a launch menu. In order to open the menu you have to hit a tiny arrow on the side of the main launcher. It doesn’t make sense really. There should really be a seperate launchmenu with a large icon.
Well, I could go on, but I have some work to do.
Perhaps you get my point?
Hi
“Also the items in the panel are positioned using a positing number that’s available in the item’s config panel. Again, what’s wrong with D&D? Not light enough?”
look at it from the implementation side. where are you dragging it from?. where would you find the app in the first place since there is no UI element yet for the new shortcut you are creating. beos did the same thing for creating shortcuts and its a dead end anyway
look at it from the implementation side. where are you dragging it from?
Where I’m dragging from isn’t important. It’s what I’m dragging that is. The panel should know the difference between a file and a panel item.
where would you find the app in the first place since there is no UI element yet for the new shortcut you are creating.
Well, how about the filemanager?
beos did the same thing for creating shortcuts and its a dead end anyway
What are you talking about? Is D&D for creating shortcuts a dead end?
The way XFCe implements it is easy from the developer’s perspective, but not from the user’s.
I’m afraid xfce inc. doesn’t hire enough full-time developers to implement everything we want in a timely manner 😉
Generating ideas is relatively easy, implementing them in a consistent and obvious way is always more difficult than you think at first.
DND for the panel has been on my TODO list from the start. It was never implemented, I’m afraid. Xfce is evolving slowly. It will go faster when more people help out.
About BeOS, calling Tracker not integrated with the system is I think underestimating its power. BeOS and the BeOS experience _is_ Tracker and yes, I think it will be very hard to create similar functionality for linux.
I know what a cipher is (most people don’t), but hell if I know what the difference between “EXP-RC2-CBC-MD5” and “EXP-RC4-MD5” is
RC2 is a block cypher (it encrypts chunks of data at a time), CBC is the chaining mode by which links the encrypted blocks together so that you can’t shuffle the blocks of data around without someone noticing. You can flip bits in the decypt in this mode but this it is noticable as you get one block of rubbish before the block with the flipped bit.
RC4 is a stream cypher, it encrypts one byte at a time and is very fast. As a stream cypher you can flip bits in the decrypt anywhere you like.
MD5 is a one way hash function, it creates a 128 bit hash of any length of data and it is very hard (practically impossible) to find two peices of data that will create the same hash. So they are used to discover if anyone has tried to monkey with the data (vital if your using RC4). MD5 is the basis for the SHA used in the US goverment digital signature standard (which is considered good enough to be used to sign laws into existance).
All where created by Prof. Ron Rivest (the R in RSA)
I’ve never heard of what EXP is.
A good analogy may be Linux is like a Patchwork Quilt that the gets a patch added to it and it evolves into something big and comfy that you’ll want to keep for ever. Beos is like a company designing an ultra lightwieght blanket that is just as warm. Not as fun to curl up in but at least you wont die from the cold.
Lets hope Santa’s little helper wont eat Linux
))))
damn im tired that made no sense
Generating ideas is relatively easy, implementing them in a consistent and obvious way is always more difficult than you think at first.
I’m fully aware of that, most of my ideas doesn’t come out 100% right when i implement them, well sometimes they come out even more right
DND for the panel has been on my TODO list from the start. It was never implemented, I’m afraid. Xfce is evolving slowly. It will go faster when more people help out.
But XFCe has been in development for how many years now? Yeah, it’s really evolving slowly. I wish I had the time and knowledge to help out. But I have too many other projects going right now. I’ll still take the time to complain once in a while though
About BeOS, calling Tracker not integrated with the system is I think underestimating its power.
No I’m not. Yes it’s part of the experience. But app_server and filesystem are much more so.
The tracker is indeed aware of filesystem attributes, but that could be done the same way in linux as well.
It’s the simplicity and elegance that made BeOS so easy to use.
Generating ideas is relatively easy, implementing them in a consistent and obvious way is always more difficult than you think at first.
I’m fully aware of that, most of my ideas doesn’t come out 100% right when i implement them, well sometimes they come out even more right
😉
DND for the panel has been on my TODO list from the start. It was never implemented, I’m afraid. Xfce is evolving slowly. It will go faster when more people help out.
But XFCe has been in development for how many years now? Yeah, it’s really evolving slowly. I wish I had the time and knowledge to help out. But I have too many other projects going right now. I’ll still take the time to complain once in a while though
For two years, maybe three, by on average 2.5 people spending a couple of hours a week. The complaining is appreciated though
It helps us set priorities.
I just wanted to say that not all usability issues are lack of understanding or lack of will, but sometimes simply a lack of time. Not that we don’t make mistakes, but you get the point.
About BeOS, calling Tracker not integrated with the system is I think underestimating its power.
No I’m not. Yes it’s part of the experience. But app_server and filesystem are much more so.
The tracker is indeed aware of filesystem attributes, but that could be done the same way in linux as well.
It’s the simplicity and elegance that made BeOS so easy to use.
Ok, yes, in fact that is what I was trying to say. Tracker derives its elegance and simplicity from the seemless integration with app_server and the filesystem and other system services.
Linux has no such consistent interface to the underlying system and therefore any simple clone of the tracker GUI will not give the same elegant experience. Developing that will be very hard, I think.
It has been said already by a few others, but his “solution” is just stupid. It doesn’t work. With simpler software, maybe, but with more complex software – the stuff actually used to get your work done – it is anathema. Imagine a new user hopping into a word processor. They set themselves as a beginning user and thus have 80% of the features hidden from them. If they are never seeing the features, how do we expect them to ever come to the realization that they are now “Intermediate” or “All”. It just adds more complexity. Thousands of people would be asking “How do I break my text in to columns?” and receiving the response “Oh, what’s your user level? OK, set it to Intermediate, then go to the Format menu and choose columns.” It’s just silly. Users don’t know what level they are in software they’ve never used. And their level of experience also varies greatly between different types of applications, meaning they would have to make these guesses well for each application. And developers would have to actually make good choices as to what the user levels should be. Bad idea.
OSS needs standards and need a centralized organization that lead its creation and implementation.
The way it is done today is really worst. I know just a few OSes and just the commercial ones are resposive and easy to use. Why there aren’t any linux software that follow the way it’s done in BeOS (that I like a lot, the only problem is it’s support for Brazilian Portuguese, both language and keyboards) or QNX, with standard API, standard user interface, etc.?
We need, as I said, a centralized organization that defines and implement standards and these standards must be followed by everyone that wants sell commercial OS based on linux.
Linux DE today seems like a lot of small pieces placed together without any integration. If we want suceed on the desktop, we need a system that has, at least, the same level of BeOS and, ideal, OS X.
These all ready exist and some of which linux it’s self does not follow well… such as not having a actual /bin/sh, most of the time…
BTW all modern OSes are like a patch work quilt… get use to it… that is where their power comes from… and with a good package system, it is absolutely nothing to worry about… BTW windows is the same way and give OSX awhile and it will be too.
Want to do something useful? Get some ppl together write a totally awesome RFC or whatever defining everything, write a lib under BSDL, existing across platform and the , along with everything else needed to make it a standard, and get it onto a few different OS(I mean OSes not linux distros)…
Remember, the world is what ever you make it to be…
BTW you are aware when you talk about stuff like this, it is not linux specific 90% or more of the time when it comes to X stuff…
<quote> I still use some KDE apps, like Quanta and KWrite, anyone have some alternatives to recommend?</quote>
I think, that all Quanta features (except WYSYWIG) has
SCREEM http://www.screem.org/news.php
It’s worth of trying.
Rain: I don’t think you meant to, but you touched another side point: Different people think different things are intuitive or simple.
(“There are no intuitive interfaces, everything has to be learned”, right?)
Also the items in the panel are positioned using a positing number that’s available in the item’s config panel. Again, what’s wrong with D&D? Not light enough?
No, what’s wrong with D&D is that it breaks the click-and-hold to open – functionality (and IMO makes it too easy to accidentally move something instead of activating it).
And what’s the deal with “Attach menu to launcher” which is the only way I’ve found to create a launch menu. In order to open the menu you have to hit a tiny arrow on the side of the main launcher. It doesn’t make sense really. There should really be a seperate launchmenu with a large icon.
Honestly,I think it is a brilliant way to both give you a list of shortcuts to your most-used programs and let you have short, concise program menues sorted by theme.
I have a “shell/file-menu” with a few file managers and shells tacked onto the aterm button, quanta and netbeans on the kate-button, kmail and knode on the firefox-button, etc.
I can see your point, though. A button where the whole button only opens the submenu can have its uses. Not that I’ve ever missed it.
Well, I could go on, but I have some work to do.
Perhaps you get my point?
Exactly
I’ll be verbose and restate my point, though:
That you don’t like it doesn’t mean neccesarily mean that it is inherently a bad interface.
I’m sure there are someone out there who likes the new XP start menu, for instance.
Rain: I don’t think you meant to, but you touched another side point: Different people think different things are intuitive or simple.
I think I meant to. Sure you are right, but providing too many ways of doing one thing isn’t really a good solution. Sometimes, forcing people into habits aren’t that bad.
One DE should focus on a certain way of doing things, while others may want to take a different approach. But trying to do everything often results in a big mess.
No, what’s wrong with D&D is that it breaks the click-and-hold to open – functionality (and IMO makes it too easy to accidentally move something instead of activating it).
You could simply have a lock/unlock feature. It’s quite common, even the XP taskbar has one. You could either rightclick and unlock or simply hold down the ctrl-key or something.
Honestly,I think it is a brilliant way to both give you a list of shortcuts to your most-used programs and let you have short, concise program menues sorted by theme.
But it’s not very logical to arrange it like that IMO. But yeah, it can be practical, but I see it as more of an advanced feature not one that most people would/could use. Besides, the little arrow button is still too small.
That you don’t like it doesn’t mean neccesarily mean that it is inherently a bad interface.
True, and I didn’t say that I rule the universe. But that doesn’t mean that there is no bad interfaces.
I think I meant to. Sure you are right, but providing too many ways of doing one thing isn’t really a good solution.
Ah, right. Anyway, that’s the point of different DEs, isn’t it? You don’t like XFCE, and gnome drives me mad.
I assume what you meant is that within one environment, there shouldn’t be too many choices?
Sometimes, forcing people into habits aren’t that bad.
One DE should focus on a certain way of doing things, while others may want to take a different approach. But trying to do everything often results in a big mess.
Exactly. Which is why I don’t agree with you on xfce.
You could simply have a lock/unlock feature. It’s quite common, even the XP taskbar has one. You could either rightclick and unlock or simply hold down the ctrl-key or something.
Ok, this one has potential.
(Though the lock/unlock button would have to go somewhere, or you would need to learn a shortcut. Nothing big, but it does add up.)
But it’s not very logical to arrange it like that IMO. But yeah, it can be practical, but I see it as more of an advanced feature not one that most people would/could use. Besides, the little arrow button is still too small.
It grew on me. It might indeed seem strange, at times, but it works, and well. For an example, consider launching a browser and a mail program:
I do the first often, so I have a firefox button. One click.
I don’t do the latter as often, so it resides under the menu from the firefox button, together with the few other internet-related programs I actually use regularly. Two clicks, and I have to scan a 5-item menu.
It helps if you like mentally grouping things, I guess.
Also, this is for me the key, defining, feature of xfce, and the one reason I use it. Yet another example of what we’re discussing.
As for the arrows, well, you can always click and hold the button. (One thing I would have liked is for that delay to be shorter.) I think they’re fine, but then I have the panel set to tiny with arrows on the side, just to save space, so I might not be representative.
True, and I didn’t say that I rule the universe. But that doesn’t mean that there is no bad interfaces.
I meant it in the least arrogant way possible
And yes, indeed.