Slashdot Mirror
Shuttleworth's Take On GNOME 3.0, Coordination with Debian
suka writes "In a fresh interview with derStandard.at, Ubuntu founder Mark Shuttleworth talks about GNOME 3.0 — its strengths, but also about what he thinks is missing. He also mentions ongoing talks for a common meta-release-cycle with Debian which could delay the next LTS."
I'm not worried about X breakages, personally. I even have an Intel 945G and I can live with the problems its causing. What I can't live with is the extreme instability of Pulse Audio. It crashes my apps contstantly from broken pipes. OK, people should be checking their pipes. But Pulse Audio itself crashes very frequently (about every hour or so on my machine). Rhythmbox won't go for more than 10 minutes without either crashing or audio failing. This is incredibly bad for me.
I realize that it's probably due to older, underpowered hardware (3 year old cheap laptop), but this should not be happening. I've yanked Pulse Audio from my machine altogether now and it's a lot more stable. I was also getting lock ups in Firefox every hour or so. Now that I've dumped Pulse Audio, I've only had one lock up in the past 3 days (still can't figure that one out -- related to video drivers???).
So, I plead with Ubuntu developers: either fix Pulse Audio, or punt it. The extra features it has is *not* worth the massive pain that some people experience.
I only hope they will follow a different path than KDE team.
They rushed to release 4.0 and since then I'm still struggling to have all the features I used to have in KDE v3.5.
And, more important, I hope that Ubuntu people won't trash GNOME v2 from night to day like they did with KDE v3.5.
Maybe Computers will never be as intelligent as Humans.
For sure they won't ever become so stupid. [VR-1988]
Ever watched someone who hasn't grown up on computers use files and folders? The physical notion may not be confusing, but the computer implementation definitely leaves a lot to be desired. I have had a 60minute discussion with someone about the distinction between copy & cut, and when it does and doesn't work. So yes... files & folders as used by computers can be enormously complex for those who are not accustomed to remembering large tree-maps ;-)
So, it finally happened! A major effort by a distro to fix one hundred really small but irritating bugs. Also known as polish. This is what Ubuntu needs, and to be fair has been quite good at. Just fixing more and more of the tiny annoyances is what creates a well-rounded desktop. On the other hand, they are introducing Gnome Shell, which while probably cool, will certainly introduce a couple of hundred new paper cuts!
https://launchpad.net/hundredpapercuts
Or you could be less elitist and realise that we're far beyond having to manually file things in this day and age, indeed that is something the computer was meant to eradicate.
A tagged document repository (with versioning history) would be best. Coupled with desktop search and changing the system file open window to be one that lets you use said search and tags to find the file instead of clicking through folders. Most files people want are more recent, so a default view of reverse chronological for the filetypes the application supports would be best.
You do, of course, still need a traditional filesystem view of this repository, and that is probably where the work will go in. Sure, tags could be folders, and you could have multiple ways of drilling down to the same file. You'd probably have a folder hierarchy that shows the most used tags at the highest level, then each subfolder is really a tag filter.
I'm not a Ubuntu advocate (more of a Fedora/RHEL person really) but if the well documented problems with KDE 4 are anything to go by then including Gnome 3 in the next LTS release of Ubuntu would be IMHO a big mistake.
An LTS release deserves to be of the highest quality from Day 1. To me it would be madness to base an LTS release on anything Gnome 3.0.
IMHO an Ubuntu LTS release whould be the desktop equivalent to RHEL or SLED in terms of stability. If it is not then you have shot yourselves in the foot. If this means being conservative in package selection then so be it.
I'd rather be riding my '63 Triumph T120.
Well, GTK+ is due for an overhaul. Fix the damn file picker. Get rid of all that excessive padding, maybe by making it themeable. Some consistency in menuitem dimensions would be nice.
Also, either give Metacity some features, at least the bare essentials, or switch to another window manager. That non-optional minimize effect is cringe worthy.
Yeah.
It would be a file library like itunes is a media library. File management would be done by the implementation instead of directly (unless you wanted to, we shouldn't take functionality away).
Some applications could use tag-discovery libraries to automate tag generation.
I'd hope it wouldn't be called "Tagged Document Repository" in the end-user documentation or presentation.
Seriously, if files and folders confuses you, you might want to reevaluate your need to use a computer.
A bit harsh, but I'd agree otherwise. I think the problem is that for those that do understand the concepts of files and directories, they balk at the idea of having to use them.
Granted it's possible that the average person in daily life has an aversion to organisation, but what I see is a relatively recent and often shrill insistence that their computer (and, by extension, the applications they use) should do their work for them and magically organise everything behind the scenes.
I consider that kind of thinking sheer laziness. And given that everyone is a system administrator (whether they like it or not), I'd suggest it's also shortsighted.
I'm suprised Shuttleworth didn't mention Zeitgeist, which is a solution to the difficulty of manually managing files and folders and is, as I understand it, being considered for inclusion in GNOME 3. The basic idea is to group files (and other activities, like web bookmarks and email contents) automatically according to human-relevant criteria, like "edited last week" or "related to this document I'm writing." It's still very much a work in progress, but it looks like it could be pretty great.
The GTK file open windows do in fact integrate desktop search, as well as a recently used file list, although the standard folder view is the default, and the search and recent options are not especially prominent. I've only recently got into the habit of using them, and they certainly are, a lot of the time, far superior to digging through some confusing mess of folders.
Feeding the troll, I know, but Google Chrome is not using Gnome or KDE because GC is designed to be a minimum functionality netbook distro, not a fully functional desktop. It may *become* a fully functional desktop, if Google is able/willing to take development that far, but whether Google's sprawling managerial structure will be able to concentrate the resources on that one project given their entrenched resource allocation tradition of "spread wide, spread thin" is something I don't think will happen in the near future.
I hate printers.
This is exactly why I hated it when MS started ramming the new term "folders" down our throats. The word "directories" suited better, as it did not conjour up invalid analogies in the minds of newbies. Now instead of just having to explain what a directory is, I now have to explain what a folder is AS WELL AS how it's not like a real folder.
I hate printers.
And then there's the bit that always gets me (and which annoys me with some of the Firefox 3 results and can be a touch annoying with too many Gnome Do extensions): They also want to to magically understand what they meant when they try to find files from the magical file system.
People have these fantastic ideas about "intelligent applications" that guess what you want, but there are just too many times when they can't/don't get it right because I know what I want and could do it quickly with a "dumb" system, but it is ambiguous to a "smart" system and so it returns excess cruft that I need to filter out.
And before anyone mentions "learning" with smart systems, that's all well and good until I want to do a different but similar behaviour, at which point it is more buried than it should be because it has "learned" that I only ever do the normal thing ;)
I've been pushing this for the last six months. I think that the best example of how to use a tagging system already exists in programs like F-Spot. A tree-like tag system goes on the left. A time-line goes on top. The files are in the main pane in reverse chronological order. Double-clicking takes the main pane into "view" mode and embeds a document, image, or video viewer. Click the "edit" button to open an editor.
The "open file" dialog in applications would be the file browser with a filter for supported files.
Put identity in the browser.
Sounds great to me - obviously you wouldn't call it that though!
Consider gmail "labels" vs traditional email/imap folders - labels are both easier to use for novices and more flexible for capable users.
YMMV, as ever.
In meatspace:
They are almost synonymous. So someone with a non-computer background won't intuitively know which one is supposed to contain which.
In computer lingo:
So it's completely unintuitive.
I think the word 'file' has its roots from the days when a 'record' was still a fundamental concept. So a 'record' is a sheet of paper, a 'file' contains a bundle of records.
I prefer 'directory'. At least then it doesn't push a false analogy on an already confused mind.
http://michaelsmith.id.au
You login, which you don't actually have to do anymore because it was too complicated, and you're presented with a fullscreen dialog box that says:
"You are too fucking stupid to use this computer. You don't understand files and folders and things. Click OK to shutdown your computer. Your computer will shutdown in 28 seconds anyway, because you're probably too stupid to work the mouse. That's the thing underneath your hand. What? That's the thing attached to your arm. Ah, fuck it. 20 seconds."
That's pretty much the entire GNOME 3.0 experience. The dialog box has been in development for the last 18 months, but obviously there's still a lot of usability testing left to do, mostly by Redhat and Canonical "engineers". The OK button logic was originally written in C but they've redone that in C# running on Mono, and Miguel de Icaza is already calling the work "superb".
Meanwhile, the KDE people have been busy readying the next batch of widgets that you will never add to your exciting K desktop experience.
Future plans for GNOME involve reducing the 3.0 dialog box down to a single pixel, then translating the status of that pixel into the power LED on your computer. This will remove the need for a display, further simplying the desktop experience and reducing enterprise costs. KDE plans to turn its entire desktop into a widget of itself, allowing you to remove it entirely with a single right-click.
Yes, my friends: the future of the Linux desktop is no more fucking Linux desktop. What a relief.
you might enjoy this article (or perhaps you've already read it?): http://unqualified-reservations.blogspot.com/2009/07/wolfram-alpha-and-hubristic-user.html
I know I did.
But they don't get hierarchies, because hierarchies don't exist in nature.
I'd have to disagree with you here. The very words we use to describe hierarchies come from nature - look at 'trunk' / 'branch' / 'leaf', 'parent' / 'child', 'master' / 'slave'. Maybe they don't instantly, intuitively get the idea when it's used as a metaphor, but that's partly vocabulary. They'll get it quickly enough if you explain to them that a 'folder' or 'directory' is a box, and a 'file' is a bit of paper that you can write on, and you can put either paper or boxes in any box.
If they don't 'get it' when it's explained that simply, then they're below the mental cutoff for that level of abstract thought. Many people (for instance) struggle to execute a sequence of simple instructions, and cannot solve even simple logic problems. They literally don't have the mental machinery required to visualise three different entities and the relationships between them, "A is next to B and B contains C". I'm not saying they're 'idiots' or that they're worthless, they just don't have abstract thought among their strengths.
Rampant carbon sequestration destroyed the Dinosaurs' tropical paradise. I'm here to help repair the damage.
In computer-space we have either directories/files or folders/documents.
In any case, word excel and powerpoint documents can contain multiple sheets of paper, and I see a lot of people take that to extremes - for example having all the day's letters contained in one word document, or every single spreadsheet they work on in one excel document.
I use that version on OS X, thanks to Fink project. While they don't promise any kind of 'final' version at this state, I can easily keep KDE 4 applications in my OS X Dock, using them instead of iTunes for example.
They are linked to actual OS X frameworks, down to Quicktime and very interestingly they use far less CPU and resources than regular OS X apps.
There are similar reports from Windows users who binary installed it and using Amarok 2 etc. right now. While on it, is there any reason why KDE 3.5 given up when KDE 4 installed? I keep using KDE 3.5 suite on OS X too. It doesn't conflict with anything at all including KDE 4.
I think what KDE 4 is and what a huge revolution it is will be understood in 1-2 years. For example when Nokia and other members of open source Symbian foundation starts using it in some form in their smart phones.
Or you could do what MacOS does to hide the unix filesystem from the user.
Arguably a Unix filesystem already is a tagged repository.
..)
In Unix-y filesystems, you don't put files in folders. You put files in the filesystem, where they get a number (inode number). Then you can set up other special files (directories) to act as indices, linking names to the inode number - as many as you want. Voila - nest-able tags (albeit not versioned in most filesystems.)
(Actually, if Unix hadn't insisted on banning '/' and NUL from filenames, a directory could in fact link arbitrary binary data to inode numbers. Bit of a missed opportunity there
Waitress: Morning! ...meatspace meatspace meatspace egg and meatspace; meatspace meatspace meatspace meatspace meatspace meatspace baked beans meatspace meatspace meatspace... ...or Lobster Thermidor a Crevette with a mornay sauce served in a Provencale manner with shallots and aubergines garnished with truffle pate, brandy and with a fried egg on top and meatspace.
Man: Well, what've you got?
Waitress: Well, there's egg and bacon; egg sausage and bacon; egg and meatspace; egg bacon and meatspace; egg bacon sausage and meatspace; meatspace bacon sausage and meatspace; meatspace egg meatspace meatspace bacon and meatspace; meatspace sausage meatspace meatspace bacon meatspace tomato and meatspace;
Vikings: meatspace meatspace meatspace meatspace...
Waitress:
Vikings: meatspace! Lovely meatspace! Lovely meatspace!
Waitress:
Wife: Have you got anything without meatspace?
Waitress: Well, there's meatspace egg sausage and meatspace, that's not got much meatspace in it.
Wife: I don't want ANY meatspace!
Those who advocate genocide deserve every protection afforded by law, and none afforded by common human decency.
A branch in real life is not part of a hierarchy. It's just a piece of wood attached to a tree; it's not contained in the tree in any way.
And no, explaining doesn't fix this problem. People know how hierarchies work; they just don't get them. They are not sure how to properly categorize things, and if they decide on a given hierarchy, later, they won't remember where they decided to put things because there's no naturally correct way of categorizing files. There's no obvious taxonomy; the problem becomes even worse when people have to use hierarchies created by other people.
The only time I've seen "normal" people successfully use hierarchies is when they created directories for years; every year got a new directory. But that's something that shouldn't require hierarchies; the OS should allow users to have temporal views on their data automatically.
As for alternatives, tagging is easier to understand since it gets rid of the whole hierarchy aspect, and since people can just add as many tags as they want to.
Actually, hierarchies really don't exist in nature. They *almost* fit how things work in nature, but every once and a while someone throws in multiple inheritance like the platypus, or someone who is both a "Student" and an "Employee", or in the family hierarchy someone will throw in a redneck or Polynesian population to gun the works. It's even worse. Sometimes, things that you thought were part of the hierarchy (e.g. in the animal kingdom, has wings versus doesn't have wings) really should be attributes since wings are "reinvented" many times in nature. Similarly, some things that used to be attributes, such as having a certain set of gene sequences at a certain locations with a certain functions, really need to be part of the hierarchy rather than the attributes. It's one reason why Ontologies are defined in declarative languages like OWL rather than a more logical hierarchal structured language.
Wouldn't surprise me if it goes back to xerox alto.
That doesn't mean that it's ultimately helpful, but it's so entrenched it seems harder tho change it than to fix it.
Oh noes. People are getting back to be being overly pedantic. Might as well start the "Untitled Folder" convo here. How can a folder be untitled yet have the title "Untitled Folder"? Because it is short for "Untitled Folder (by you)" you clods. Now that is out of the way...I shall hide under my rock again.
Once you start despising the jerks, you become one.
Just to add a bit more information... one way I think of this is in terms of native languages. Most people who learn a second language, always need to translate through their primary language. E.g., if I learnt chinese, it might go: input in chinese => translate to english => concoct reply in english => translate to chinese => output in chinese
This is a lot of overhead. Compare this to a native chinese-speaker... who simply hears in chinese, and thinks & responds in the same language.
People who didn't grow up with computers don't develop the ability to "think" in computer terms like those of us "native" speakers. As such, dealing with files & folders, they need to go through this long translation process to a real-world analogy and back.
The problem is, the files & folders analogy is very thin. Have you ever lost a hand-written letter because the power suddenly went out?! Or mis-placed a single document in a huge huge stack of papers and photographs (e.g., more than 200). The answer to both is probably not... because a stack of paper 200 documents long is unmanageable in the real world... and a hand-written letter autosaves every change you make ;-).
So yer, from this, a few breaks between the user's expectations of how folders and files work, and pretty soon it seems like a mystical cave. The user doesn't remember the exact folder sequence (after all, you can layer them ridiculously deep), and forgets where things are saved, and pretty soon they create a gnarly mess of their files & folders... and are lost!
Something that is perhaps more predictable is the idea of time-based or activity-based files (without folders). Gmail tags I find are also far more useful (means the user can simply search as they think of it, such as computer > essay > 2009 > university, or 2009 > university ...etc.)
Either way, I would highly recommend observing some beginner users discretly if you can... I feel strongly it has helped me better understand how my users may see the programs I write :)
MS-DOS has DIR command, not FOL. Can't blame the early DOS jockeys for this though, cause they just borrowed the convention from VMS.
At least this is one thing that MS, DEC and Unix can all agree on: "directory" is correct, and "folder" is dumb.
In computer-space we have either directories/files or folders/documents.
MS Windows uses "Files" and "Folders". Actually, the type of a folder is "File Folder". A "document" is a subtype of "file", because a "file" can also be an "application" or "program". A "file" could also be an "archive", which contains "files" and "folders". Some "archives" are called "cabinets". And of course all these files, folders, archives, cabinets, etc. can be seen in windows. Of course to be able to see those windows you first need to use a key, to open the lock. And to add protection from outsiders to see you files, etc. we have walls of fire. All the files, folders, etc. are stored on something called "drives"... why on earth are they not called "rooms"? And I wonder where the "roof" is in all this stuff.
I don't think so. How many times have you heard someone say: "Put that paper in the file" or indeed, "Put those papers in the folder".
You'll find people are more likely to say: "Do you have those files?", or "What's in that folder?".
A file was always a ordered collection of papers/photos/data on a singular topic. A folder was always a place where you put papers and files. The file/folder analogy in computing is actually a very good one, and the only people who are ever confused by it are the find of people who would be confused by real files and folders anyway.
The problem isn't actually with the system. It's actually with users who cannot and will not adopt any method or organisation over their own files. Admittedly, the default folders most programs obnoxiously set complicates things, but the proof of the pudding is when you ask someone where their files are and they give you a helpless stare. Sometimes they have been using computers, and these very files, for years. yet they have absolutely no idea what a file is, where their files are, or even of their existence outside of the context of the exact program that manipulates them.
As they proceed to recite the unique set of incantations they use to access a particular file type, you begin to see why "cloud computing" and web based software is not going to be a huge leap for these users.
May the Maths Be with you!
In any case, word excel and powerpoint documents can contain multiple sheets of paper, and I see a lot of people take that to extremes - for example having all the day's letters contained in one word document, or every single spreadsheet they work on in one excel document.
During the late 80s/early 90s I worked for a firm that had a satellite office with a single PC which was running Wordpress on DOS. The secretary there had a single document containing every single letter she had typed over the past three years. She typed letters for an office of 15 engineers and regularly wrote several every day.
Worse still, when she opened it (fortunately just the once per day) she would press the down cursor key repeatedly until she got to the last line. She spent approximately half an hour doing this I asked her how she found an old letter to check, and she replied it would be in the filing cabinet behind her. No matter ho many times I tried to show her how to use individual files, she went back to this single document. I once discovered we had no backup of this single file (it was saved outside of the document directory) and I still have the occasional nightmares about it.
I think more people would expect it to be a wall of fire.
Sooner or later, everyone re-invents VMS.
I don't think so. How many times have you heard someone say: "Put that paper in the file" or indeed, "Put those papers in the folder".
You'll find people are more likely to say: "Do you have those files?", or "What's in that folder?".
We *may* have stumbled on a US vs UK issue here. Where the rest of the world sits, who knows?
Definitely in the UK, you would go to a stationer's and buy a "file" in which to keep paperwork. "Put that paper in a file" is something you'd expect to hear, and is exactly equivalent to "Put that paper in a folder".
It certainly would be nice if only people qualified to use a computer did so, but it won't happen any time soon.
Would it be nice? I don't think it would. If only people who are "qualified" to use a computer did so, we'd still be paying $3000 for Pentium I-era technology. The vast majority of computer purchasers buy their computers to do something else with them, not to become computer experts. It's that large market that allows for the economies of scale that drive processing power up, and price down. Without those "unqualified" computer users, computing would still be an expensive hobby.
Rule of Slashdot #0: You and people like you are not representative of the larger population. - A.C.
The problem isn't actually with the system. It's actually with users who cannot and will not adopt any method or organisation over their own files. Admittedly, the default folders most programs obnoxiously set complicates things, but the proof of the pudding is when you ask someone where their files are and they give you a helpless stare. Sometimes they have been using computers, and these very files, for years. yet they have absolutely no idea what a file is, where their files are, or even of their existence outside of the context of the exact program that manipulates them.
This is true, and in a way the "File->Save" UI reinforces it. You type in a name, you don't think about where the CWD is, and when you "File->Open..." all your documents are there.
Sun's old OpenWindows had a nice UI which might have helped with this. The text editor app had an icon in the corner representing the file you had open. To save, you dragged that icon to a file window. To open, you dragged an icon from the file window on to that part of the text editor. It means your save is not separate from your filesystem.
Linus did.
Maxim: People cannot follow directions.
Increases in truth directly with the length of time spent explaining them
http://live.gnome.org/GnomeZeitgeist
"When I first heard Daydream Nation it quite frankly scared the living shit out of me." -- Matthew Stearns
In any case, word excel and powerpoint documents can contain multiple sheets of paper, and I see a lot of people take that to extremes - for example having all the day's letters contained in one word document, or every single spreadsheet they work on in one excel document.
And for those types, I recommend a clue bat be applied, liberally until they get the message that one mistake and the whole day is gone. Sheesh. Makes me more FOR a license to run a computer all the time. If they are so fond of a format they've cobbled up, then save the SOB as a template so they don't have to feel like they will have to re-invent the wheel to do their next proposal. That 'creation' drudgery may be putting a lot of the dummies off from doing it right.
--
Cheers, Gene
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
The NRA is offering FREE Associate memberships to anyone who wants them.
Seize the day, put no trust in the morrow!
-- Quintus Horatius Flaccus (Horace)
I think the word 'file' has its roots from the days when a 'record' was still a fundamental concept. So a 'record' is a sheet of paper, a 'file' contains a bundle of records.
This makes sense, some historical information can be found by looking at the ASCII control character assignments. Look at the end of the control characters, in reverse order they were supposed to be larger and larger block separators:
SPACE (word separator)
US (unit separator, like between columns in a table)
RS (record separator, a row in a table)
GS (group separator, equivalent to an XML grouping, but they never thought of hierarchy?)
FS (file separator)
I think the figuring was that any larger grouping than "bunch of files" would be "a different tape" so there was no need for a larger separator.
I think it would be nice to reuse these control characters for markup so we don't have to worry about escaping things all the time, too.
I definitely agree that the tree-hierarchy has it's place in the computing world. As you said, there are some things that are so solidly orthogonal that you want to restrict them to only one folder.
But I don't believe that forcing everyone through the same system is helpful. I know many things I have that simply don't fit well in folders. My music collection for example... how do I sort this? If I go by artist, it doesn't hold collaboration volumes very well, if I go by album, it is difficult to remember the artist... If I want to go by genre, I must choose only one, and this makes categorization of some artists very difficult where they seem to entirely change style over the course of a single album.
What I am saying is that hierarchical tree storage has it's place, but I don't think it is the most intuitive way of storing things :)
I agree that search-only functionality sucks btw. I even went so far as to make an interface in java that behaves like a folder interface, only it uses a tag system. In this view, a sub-folder is a tag that isn't already in the hierarchy, that is present on one or more of the child items. Allows folder navigation for those who like folders, addresses your point about easily seeing parent levels if desired... but also allows a file to live in multiple hierarchies... it worked nice. Not sure if you could base an OS on it though xD