As for the Alexandria updates (assuming Alexandria is the SF codebase), I just don't see any reason to think anything other than the fact that Alexandria is more likely to be purchased by companies if it isn't freely available as an alternative.
I am a hot chick with a geek fetish who desperately wants to get laid. Please explain to me how I might manage to do this...
I think the problem is that you aren't talking to usability folks -- most OSS projects don't currently have a "usability guy" or often even a designated "interface programming guy". It's kinda too bad. I'd like to suggest that if you are interested in GNOME, try the Usability Project folks, who *are* interested in taking and cataloging suggestions. Dropping by their IRC channel and working on identifying user interface issues with GNOME is something that a lot of folks would like. Sun donated some UI people's time to the project for a decently-sized study, but they always would like more input. If you like KDE, try the KDE Usability Project. Since these two are about the most usability-focused projects around, they're probably a good place to put in some improvements, where your work won't get dropped or ignored.
One of the main problems with OSS is that generally programmers are used to, y'know, doing their own thing. It's hard to take advice from someone else on a volunteer project, especially since they might get less of that precious credit for an improvement (and credit must be a primary motivating factor, since the code-for-self motivation generally isn't when it comes to UI). Hence the "well, why don't *you* implement it" business.
I realize that funding is a problem when it comes to doing studies, but you probably *do* have a knowledge of existing problems, stuff that people like Apple have run into in the past (I read interface articles as a bit of a hobby, and I've got a ton of interesting things that people have run into to try to avoid).
Do remember, though -- usability people and artists are the resource in shortest supply in the OSS world.
Here's motovation for you -- criticism. As part of the motovation in Open Source is recognition, the last thing you want is your project to be seen as "This is why you shouldn't use Open Source". On my project I hold everyone to a high standard of quality....I generally don't like criticism, just because it discourages people from volunteering their time.
You do Lightbringer? There are almost no good, sizeable, complete FLOSS games in existence. There are *many*, *many* open-source games that have been started but will never be finished. There are very few sizeable games. Even huge projects like WorldForge, with many, many content producers and coders *still* haven't finished. Based on what I've seen, I would estimate that your chances of finishing Lightbringer at being less than 1%. Just about every young developer I can think of has, at one point or another, considered writing a game of their own. You would be much better off contributing effort to an existing project.
That was criticism. It was constructive (encouraging you to move effort to a place that it will likely do more good), but it was also very discouraging, and I would generally want to avoid handing out such criticism out on an OSS project. If someone writes a feature, it's a *pain* to be told that it isn't going to go in, especially if they use that feature, like it, and have documented it, tested it, and bounced it off of other users. Sure, a few projects can afford to turn away features because they have developers breaking the door down. There is no shortage of people wanting to add features to the Linux kernel, for instance. But most projects can't afford to discourage developers much.
Use Knoppix. It's what I recommend to anyone that wants to dabble with Linux and see what it's all about. You don't have to worry about dropping a bootloader on your system or in repartitioning. Just burn a CD, set your BIOS to check the CD drive before the hard drive in the boot process (if it isn't already), drop the CD into the drive, and reboot. Yeah, it isn't quite as fast (due to having to hit the Cd instead of a hard drive) as a traditional Linux install, and you don't have lots of hard drive space to play with, but it's a good way to take a look at what's available.
Open Source tends to be full of language and tools bigots. Use Eiffel for Gnome instead of C! Use Python instead of PHP! Use Perl instead of Shell! Use Lisp for everything! Don't use Mono for anything!
They aren't bigots. They are advocates. What you are seeing is a healthy, thriving compiler ecosystem, with new languages and ideas struggling for dominance and survival. Microsoft killed most other languages by choosing to support a very specific set of languages in Windows (C/Win32, C++/MFC, C#/.NET, VB/scripting), and crushed all the other Windows compiler providers, and there are no alternate compiler vendors because Microsoft crushed them one by one. It reminds me of the pretty, crystal clear pools you see after chemical runoff, where there are no organisms because everything is dead. Without competition, there is little reason for improvement.
Why (to name a case that's close to my heart) is MSIE stagnating and still lacks support for PNG alpha? Because everything else is *dead* and it has no serious competition.
I can't agree. Your argument is made based on a basic thesis that the OS X interface is *better* than current FLOSS GUIs. It has a much more shallow learning curve than, say, sawfish, but lacks when it comes to user power.
Let me give an example. I use sawfish. A major problem with click-to-focus windowed interfaces dating back to at least classic Mac OS is that of the "wake up click". When you click to bring a window to the foreground and grant it focus, you have two choices as a UI environment designer -- either the click is not passed through, or it is passed through. If you do not pass the click through, and have a number of non-overlapping windows on-screen and are doing work involving multiple windows (for example, with Photoshop in a document and palettes), every action involving switching windows will require an additional click. Plus, the user must be especially aware of which window is in the foreground, or else he will double-click in the active window instead of single-clicking in an inactive window. The second approach, which is used by all GUIs that I can think of offhand, is to pass through the click. This means that whenever the user is working with multiple overlapped windows and wants to bring one to the front with the mouse, he has to identify an aread that does not contain an active widget. This is, frankly, a pain in the ass and leads to accidental clicks occasionally. (It also confused the dickens out of users in the days of classic Mac OS, which used a hybrid approach -- the wakeup click to switch applications was *not* passed through, but a click to switch windows within an application was -- usually, except for a couple of apps like the Finder -- not passed through.)
I can code in lisp (a bit, enough to make sawfish dance the way I like) and happen to have a five-button mouse. I have my fourth mouse button, when clicked on a window, bring that window to the foreground, focus it, and begin dragging it. As a result, I don't need to look for a titlebar to bring a window to the front or to begin moving that window.
Apple does not have interest in this kind of functionality. Heck, we can't even convince them to add a *second* button to their mice -- functionality relating to four buttons is clearly not something they want to poke at. It is not a huge deal for Joe User, because he has that one-button mouse. For me, it *is* a significant issue. I cannot add such a feature to Mac OS X.
Take Expose. It's certainly a pretty interface, and it is probably the best interface I've seen for managing a single-viewport system. It is not, however, nearly as powerful as a multiple-viewport system, even though it's less confusing to Joe User ("Why did all my windows vanish suddenly?"). Apple is not interested in providing multiple viewports with OS X, because it's not a feature that will make them much money -- it isn't demanded by the masses, and it doesn't make a sexy demo for them.
The thing that commercial organizations do is *target the masses*, because that's generally where the money is. OSS developers may well not do that -- for them, the main priority is to make something that *they* want to use. Sure, fame is nice and all that, but most folks that use a piece of software don't want to work on it very much -- if a maintainer finds a better piece of software, his piece of software generally goes away.
As a result, the fact that Joe User may not understand what "backing store" is and the configuration for such an option might confuse him takes a back seat to the fact that the developer can squeeze double the performance out of his system by adding such a feature.
Joe User *may*, if one is lucky, be comfortable with globbing. He probably does not understand regular expressions, but generally regular expressions allow a knowledgeable UNIX user a tremendous amount of power. Should applications omit regex searching, since the menu option for it might be confusing? If the goal is to recruit Joe User, the answer is probably yes. If the goal is to make the best possible system for techies, the answer is no.
It's not that it isn't a problem, it's just that the title of the paper implied that this was an Open-Source-specific set of problems. It's like saying in a car review "This car model has a significant problem -- it broke down after 250,000 miles. This lack of reliability has caused me a good deal of inconvenience." While this may well be true, the fact that *most* cars are likely to break down at or before 250,000 miles is a reasonable thing to point out. People don't want the criticism to be taken out of context, and to make a bogus decision based on that criticism.
Not only is it less glamorous, it's less fun and users complain about UI. I followed the gtk-gnutella mailing list when the UI was redesigned, and I remember how much complaining the UI guy got with each change he made.
This is not the case when you hit some obscure RPM bug.
What bug did you hit in RPM?
nope. It's UI-centric.
What?
I'm not sure I understand what you mean.
I would say that Office is indeed feature-centric. As a matter of fact, the feature-centric argument is the one argument that I don't agree with at *all* in the article. Commercial software needs bullet points to put on the box. Refactoring source code is hard to justify. I see people cleaning up code *all the time* in the open source world.
_THAT_ is the authors point exactly.
I dunno. I think the "design for self" issue is a valid complaint, but that it isn't generally *that* awful in practice. For instance, Samba isn't all that user-friendly to someone used to a GUI Windows environment -- the authors designed to to be efficient and powerful, but weren't worried about trying to bundle a GUI. It just means that other people make front ends.
I still cannot get my network printers working properly with CUPs.
That's really, really funny. I heard the bitter complaining for ages about CUPS -- how hard it is to configure, how many problems it causes. Yesterday, *I* found myself finally needed to set up printing on a box that had CUPS on it rather than LPRng or something similar.
I ran system-configure-printer. A GUI interface came up that asked me what I wanted to call the print queue. Well, I guess that I admit that I know that it's okay to call a print queue whatever I like from UNIX days, but I think that the dialog wasn't too confusing about making up a name, and listed all the constraints on me. It let me enter an optional description.
On the next screen, I finally managed to figure out why the CUPS configuration seems to confuse so many people. When you click "help", instead of giving you "help" on the entire dialog as I had expected, help for just the currently-selected menu option comes up -- LPD, CUPS, locally-connected, etc. That was unexpected on my part. However, I happend to choose the one I wanted to know about before hitting help, and so didn't have any problems (until I went back to examine the interface for UI issues just now). I remembered ESR getting CUPS and LPD confused, so I hit "help" when I happened to have CUPS up, and realized that I wanted LPD. I chose LPD.
I was given the option to enter an IP and queue name. I know the IP of the printer, and I don't think that it's all that exceptional to need to enter an IP -- any system being set up these days is going to need at least a *couple* of numbers or some basic configuration done when being set up. The queue name threw me for a minute, so I clicked "help" and was informed that the right option is usually "lp". I guess it would have been nice to have the default value filled in, but I don't think anything on this screen was *that* hard to manage.
I was then asked to select the printer type. The documentation said that "many people may have to select Generic Postscript", making it sound like a sub-par option, when it was, in fact, exactly what I wanted. I guess that was a bit confusing. It might be better, since this is just a high-level GUI config utility, to try to maintain a database of vendors and models, and choose the appropriate setting, or perhaps have an "unlisted" entry.
Finally, the thing told me that it was going to print a test page. I clicked OK, and a test page spat out of the printer. Seemed to work fine to me. I can't see what's so hard about CUPS, really. I've had *much* more complex problems trying to convince Windows to do something (and, admittedly, with other Linux software) than with CUPS.
The big advantage of open soure - sharing and code reuse - just isn't happening. For example, I'm looking for a new OSS webmail program. I have certain criteria. There are something like 209 projects related to this on Freshmeat. Why? That is no more efficient than a couple hundred closed-source vendors doing separate implementations of competing products. What have we gained?
Just out of curiosity, were you able to find a webmail package that you liked?
Is this an open-source-specific issue? It seems to me that there were a tremendous number of webmail packages, both open source and closed source, produced during the Internet boom. The difference is that a lot of the proprietary webmail vendors have folded, but the source to the open source projects still is available for anyone who would like to adopt.
Honestly, I don't think Freshmeat is the greatest system in the world for recommending a good "currently-best-available" software package. It really isn't intended to do so -- it's an attempt to be a complete catalog. I will grant that the Open Source world could really use a "this is what I currently consider best to do X, and this would be my second choice" website.
2. Choice really isn't, sometimes. When I am faced with two OSS projects that supposedly perform the same task, I often find that each lacks a function that the other has, and neither do everything they should. Why don't they team up and make a single, solid project?
For the same reason that all countries don't get together and make one world government. Robustness is important, and it's generally best gotten by having a number of people taking their own approach. Suppose one team is dead set on C++, and another thinks that servers are better written in Ocaml -- there's a pretty fundamental divide there, which can only be resolved by the creation of two projects. The Ocaml team thinks that the use of C++ will lead to too many security flaws, and the C++ team thinks that the use of Ocaml will lead to not enough coders and hence not enough functionality. Either one might be right -- the only way to find out is to start implementing. I'm glad that these different approaches exist -- without this, we'd all be out a lot.
Also, one of the tasks of a maintainer is when to say that a feature should go and when to say that a feature should say. This role is crucial for any kind of disciplined development, and is necessary even in the more relaxed open source world. Suppose you *really* want a feature in KDE but the KDE maintainer tells you that your idea is bad, and that you should piss off. You can then add your idea to GNOME, and see what happens. If your idea works out, eventually it's likely that KDE will adopt it as well.
The more Windows-like Linux becomes, the less I am interested in running it. I've mostly switched to FreeBSD (I used to be Redhat only).
I use Fedora, and I'm curious as to what elements of RH bother you.
I use sawfish, xbindkeys, and gkrellm for a desktop. I use xemacs and a bit of vi. Sure, the default environment is GNOME, but it's not as if you're constrained to use just GNOME.
I do agree that there are some things that irritate me that have been done in the name of ease of use.
For example, GTK/GNOME 2 both have user-rebindable accelerators disabled by default, and require config file modifications to re-enable them. This was done strictly to assist newbie users, and represents the removal of extremely powerful functionality that advanced users must hunt around to re-enable.
I've had cell phones for about 6 years now and they have all stayed off, why? because I'm more important than any other jerk that wants to ring me, there is hardly any good reason to use them, if somebody wants to reach me email is the preferred method, I'll check it several times a day and I don't have to listen to their endless babble.
I used to think roughly that (with the landline) and just didn't answer the phone, and referred people to email, checking the answering machine once a week or so. That was before I decided that *that* was too generous. I never read my email more than once a day, and after it kicked up in volume I started handling it about once a week to once every couple of weeks. The people that *really* know me can reach me via instant messaging protocols when I'm in front of my computer, and my employer can reach me at work (as can people with work-related communications). Other than that, I'm not a butler, and I don't see why masses of people all around the world should be able to ring me on demand (especially advertisers). My time is my time -- not time to be spent at the beck and call of others.
Plus, if people find that they can't immediately contact you, they have a tendancy to try to resolve problems themselves -- normally it's just "easier" to ask someone else every five seconds, which ends up overall wasting time (from communication overhead and context switching overhead).
Really? What makes you think OSDN won't start charging for SourceForge.net in the future? If so, what's your plan to migrate your projects to another server? If the SourceForge code were available, anybody would have the Freedom to start up a successor. That's what GForge is for.
Don't you view that as a sort of paranoid approach?
I mean, sure, if GForge works as well as SF, then it might be a good choice. But slamming SF because you think that they're suddenly going to clamp down on all the data they serve, without any evidence to support such a claim...that's paranoid. I recognize that there is a risk, but there's risks everywhere -- it's essentially impossible to eliminate everything. What if the FSF goes bad? Perhaps more plausibly, what if Stallman has a heart attack and a corporation manages to gain control (via bribery or whatnot) of the FSF, and hence has the freedom to write the GPLv3, with which GPL software by default can be automatically moved to. There was some discussion of giving special privileges to GPL-friendly companies, which IIRC what started Linus on his "I'm releasing my software under GPLv2 only" kick. The GPL controls a phenomenal amount of IP, probably more than any other legal document in history -- the ability to affect that document would be worth almost anything.
We have no way of knowing for certain who will turn out to be a baddie. We just try to generally reduce risk and react as things happen. Thus far, I've been pleased with the services provided by SourceForge, and have no reason to think that they will change their stance in the future.
The only difference between nongnu.org projects and gnu.org projects is that Free Software Foundation Inc owns the copyright in the source code of gnu.org projects.
Sure, but given all the domain names in the world, Savanah chose to use the domain name "nongnu.org", and proposes that software projects on SourceForge move there. Frankly, that's a political message that they propose that an awful lot of people should look at each day.
For example, GCC is under constant pressure by RMS to move from its own server (that happens to be hosted at Red Hat) and onto Savannah.
I really love it when people criticise Red Hat as not being oriented enough toward the community, and ignore the constant hosting, patches, technical work, donations, and political oomph and Red Hat provides to the open source world.
I wouldn't work on an Open Source project that required me to pay to work on it. It's just not reasonable.
I can understand them providing additional services, like POP3 email access @sourceforge rather than just email forwarding, or something like that, for money. However, if SF tries doing something like this, they are, simply and plainly, going to go away.
Savannah is lesser used -- there are fewer adherents of Free Software than Open Source.
The Open Source stance (as exemplified by ESR) is a more pragmatic one than an ideological one -- that people should use Open Source rather than Free Software because it *works better* than closed source, not because of a moral or philosophical mandate. The primary issue that SourceForge detractors bring up is that the current codebase is not available; this is an issue to a number of people strongly ideologically aligned with Free software, who want to interact with nothing but Free software. There is a parallel here. Since SF costs nothing, works well, and helps spread and facilitate open source software, there are few pragmatic issues with SourceForge that Savannah solves. Thus, the issues with Open Source that Free advocates have are mostly the same complaints that are raised about SourceForge.
Savannah's main issues are caused by a lack of people working on it, and it is currently less ready-to-go than SourceForge. It's HURD and Linux in a mirror.
Savannah makes its feelings on the importance of Free software very clear with the nongnu and gnu names. The SF people don't particularly place a lot of emphasis on someone being associated with a project or having a particular license -- there's no sourceforge.sortaopen.net for BSD-licensed projects, for instance.
Finally, while this is more germane to this story than to SF in general, the politics in the linked-to story remind me a good deal of the complex and never-ending debates about Free software purity that come up more frequently in the Free Software world.
I suppose that a lot of Free advocates are going to view this as a bit flamish -- I guess it's a bit cutting in that it identifies that Savannah hasn't been operating as well as SourceForge, but I don't feel that it's particularly false or misleading.
I use the GNU utilities as well as Apache every day -- I like both chunks of software.
I also, as people who read my posts frequently know, tend to often feel a bit frusterated with Free advocates. I do, not infrequently, think that Free folks can come off as a bit too rabid to the general public -- this mainly becomes an issue when media, desperate for some kind of figurehead for the open source world, settle on RMS, and he propagates his (intimidating to a CTO) views on intellectual property. I also remember when the Crystal Space team (an excellent LGPLed 3d engine), wanted to be absolutely correct WRT the GPL and valuing Stallman's input, wrote him to ask for a bit of clarification on a licensing detail. Stallman's response, an enlightening read, highlights a good deal of what I consider the difference between Open Source folks like Jorrit and Free folks like Stallman.
For one thing, people may not realise just how annoying they're being; this lets them know, and gives them a chance to improve their manners.
If they are willing to go out of their way to be polite. The current solution *is* to ask people to behave differently, and it's clearly not working. Oh, it's fine for the rude users, but it works much less well for those people who have to put up with it.
Furthermore, it requires continual effort on the part of the people being imposed upon -- having to go up and hassle with some stranger on the subway each day is not anyone's idea of a good morning.
A jammer may have be shiny and give you a feeling of power, but it's still a bad solution to the problem.
"shiny and a feeling of power, but a bad solution to the problem it tries to solve"...that description is terribly ironic, as I would have used almost the same criticism, word for word, about cell phones and useful communication.
AFAIC, a loud and annoying mobile phone ring is no better or worse than an equally loud and annoying digital watch tune, and a loud and annoying conversation on a mobile phone is no better or worse than an equally loud and annoying conversation of any other kind.
Yes, but:
(a) most people don't use digital watch alarms constantly -- many bored people call other people to socialize all day long.
(b) digital watch alarms aren't as loud, and are simple and easy to tune out.
(c) the conversation, as the story pointed out, is much worse than the ring tone.
I just happen to feel that generally, technical problems are better than social ones, and require everyone involved to do less work.
Instead of expecting every person being annoying to have to try to convince an irritating cell phone user to stop, the phone conversation is simply jammed. This shifts the onus of putting out effort from the people who don't want to be annoyed by phones onto phone users to be less annoying and companies to sell less annoying products. You don't want to be jammed, speak quietly, politely, and use a vibrator.
Just as I don't think it's feasible for cell phone opponents to fight technology -- cell phones are going to be here and increasing in use, and people are going to be assholes with them, I don't think it's feasible for cell phone advocates to fight technology -- a jammer need not be very large or powerful, and is essentially impossible to effectively track down.
This is the equivalent of posting, "My AMD processor is better than your Intel processor!" It's a quote designed to ignite a fact-less argument on who has bigger ones.
No, no. That *would* have content -- it would mean that some consumer, for one reason or another, is willing to back a purchase he made. *This* is contentless -- an exec at a company says that really, his products (which are being squashed in the marketplace) are better than the competition. You'd expect that from any company exec.
Slashdot Mirror
As for the Alexandria updates (assuming Alexandria is the SF codebase), I just don't see any reason to think anything other than the fact that Alexandria is more likely to be purchased by companies if it isn't freely available as an alternative.
They wouldn't need to -- that's why they'd be taking control via something like bribery, not purchasing outright.
Usability and UI are exactly my background. Please explain to me how I can have a positive impact on OSS...
Wow. This reads something like:
I am a hot chick with a geek fetish who desperately wants to get laid. Please explain to me how I might manage to do this...
I think the problem is that you aren't talking to usability folks -- most OSS projects don't currently have a "usability guy" or often even a designated "interface programming guy". It's kinda too bad. I'd like to suggest that if you are interested in GNOME, try the Usability Project folks, who *are* interested in taking and cataloging suggestions. Dropping by their IRC channel and working on identifying user interface issues with GNOME is something that a lot of folks would like. Sun donated some UI people's time to the project for a decently-sized study, but they always would like more input. If you like KDE, try the KDE Usability Project. Since these two are about the most usability-focused projects around, they're probably a good place to put in some improvements, where your work won't get dropped or ignored.
One of the main problems with OSS is that generally programmers are used to, y'know, doing their own thing. It's hard to take advice from someone else on a volunteer project, especially since they might get less of that precious credit for an improvement (and credit must be a primary motivating factor, since the code-for-self motivation generally isn't when it comes to UI). Hence the "well, why don't *you* implement it" business.
I realize that funding is a problem when it comes to doing studies, but you probably *do* have a knowledge of existing problems, stuff that people like Apple have run into in the past (I read interface articles as a bit of a hobby, and I've got a ton of interesting things that people have run into to try to avoid).
Do remember, though -- usability people and artists are the resource in shortest supply in the OSS world.
I would like to refer you to the Interface Hall of Shame. Please read an interfacq critique or two, and reconsider just slapping those buttons in.
Here's motovation for you -- criticism. As part of the motovation in Open Source is recognition, the last thing you want is your project to be seen as "This is why you shouldn't use Open Source". On my project I hold everyone to a high standard of quality. ...I generally don't like criticism, just because it discourages people from volunteering their time.
You do Lightbringer? There are almost no good, sizeable, complete FLOSS games in existence. There are *many*, *many* open-source games that have been started but will never be finished. There are very few sizeable games. Even huge projects like WorldForge, with many, many content producers and coders *still* haven't finished. Based on what I've seen, I would estimate that your chances of finishing Lightbringer at being less than 1%. Just about every young developer I can think of has, at one point or another, considered writing a game of their own. You would be much better off contributing effort to an existing project.
That was criticism. It was constructive (encouraging you to move effort to a place that it will likely do more good), but it was also very discouraging, and I would generally want to avoid handing out such criticism out on an OSS project. If someone writes a feature, it's a *pain* to be told that it isn't going to go in, especially if they use that feature, like it, and have documented it, tested it, and bounced it off of other users. Sure, a few projects can afford to turn away features because they have developers breaking the door down. There is no shortage of people wanting to add features to the Linux kernel, for instance. But most projects can't afford to discourage developers much.
No.
The problem is that Microsoft is a *monopoly*.
If BeOS or Mac OS X wants to be proprietary, most people don't *care* because they aren't forced to interact with it or be compatible with it.
With Microsoft, you don't get a choice. You have to deal with Windows, you have to deal with Microsoft file formats and protocols, end of story.
You can always legitimately complain about an artificial monopoly.
Use Knoppix. It's what I recommend to anyone that wants to dabble with Linux and see what it's all about. You don't have to worry about dropping a bootloader on your system or in repartitioning. Just burn a CD, set your BIOS to check the CD drive before the hard drive in the boot process (if it isn't already), drop the CD into the drive, and reboot. Yeah, it isn't quite as fast (due to having to hit the Cd instead of a hard drive) as a traditional Linux install, and you don't have lots of hard drive space to play with, but it's a good way to take a look at what's available.
Open Source tends to be full of language and tools bigots. Use Eiffel for Gnome instead of C! Use Python instead of PHP! Use Perl instead of Shell! Use Lisp for everything! Don't use Mono for anything!
They aren't bigots. They are advocates. What you are seeing is a healthy, thriving compiler ecosystem, with new languages and ideas struggling for dominance and survival. Microsoft killed most other languages by choosing to support a very specific set of languages in Windows (C/Win32, C++/MFC, C#/.NET, VB/scripting), and crushed all the other Windows compiler providers, and there are no alternate compiler vendors because Microsoft crushed them one by one. It reminds me of the pretty, crystal clear pools you see after chemical runoff, where there are no organisms because everything is dead. Without competition, there is little reason for improvement.
Why (to name a case that's close to my heart) is MSIE stagnating and still lacks support for PNG alpha? Because everything else is *dead* and it has no serious competition.
I can't agree. Your argument is made based on a basic thesis that the OS X interface is *better* than current FLOSS GUIs. It has a much more shallow learning curve than, say, sawfish, but lacks when it comes to user power.
Let me give an example. I use sawfish. A major problem with click-to-focus windowed interfaces dating back to at least classic Mac OS is that of the "wake up click". When you click to bring a window to the foreground and grant it focus, you have two choices as a UI environment designer -- either the click is not passed through, or it is passed through. If you do not pass the click through, and have a number of non-overlapping windows on-screen and are doing work involving multiple windows (for example, with Photoshop in a document and palettes), every action involving switching windows will require an additional click. Plus, the user must be especially aware of which window is in the foreground, or else he will double-click in the active window instead of single-clicking in an inactive window. The second approach, which is used by all GUIs that I can think of offhand, is to pass through the click. This means that whenever the user is working with multiple overlapped windows and wants to bring one to the front with the mouse, he has to identify an aread that does not contain an active widget. This is, frankly, a pain in the ass and leads to accidental clicks occasionally. (It also confused the dickens out of users in the days of classic Mac OS, which used a hybrid approach -- the wakeup click to switch applications was *not* passed through, but a click to switch windows within an application was -- usually, except for a couple of apps like the Finder -- not passed through.)
I can code in lisp (a bit, enough to make sawfish dance the way I like) and happen to have a five-button mouse. I have my fourth mouse button, when clicked on a window, bring that window to the foreground, focus it, and begin dragging it. As a result, I don't need to look for a titlebar to bring a window to the front or to begin moving that window.
Apple does not have interest in this kind of functionality. Heck, we can't even convince them to add a *second* button to their mice -- functionality relating to four buttons is clearly not something they want to poke at. It is not a huge deal for Joe User, because he has that one-button mouse. For me, it *is* a significant issue. I cannot add such a feature to Mac OS X.
Take Expose. It's certainly a pretty interface, and it is probably the best interface I've seen for managing a single-viewport system. It is not, however, nearly as powerful as a multiple-viewport system, even though it's less confusing to Joe User ("Why did all my windows vanish suddenly?"). Apple is not interested in providing multiple viewports with OS X, because it's not a feature that will make them much money -- it isn't demanded by the masses, and it doesn't make a sexy demo for them.
The thing that commercial organizations do is *target the masses*, because that's generally where the money is. OSS developers may well not do that -- for them, the main priority is to make something that *they* want to use. Sure, fame is nice and all that, but most folks that use a piece of software don't want to work on it very much -- if a maintainer finds a better piece of software, his piece of software generally goes away.
As a result, the fact that Joe User may not understand what "backing store" is and the configuration for such an option might confuse him takes a back seat to the fact that the developer can squeeze double the performance out of his system by adding such a feature.
Joe User *may*, if one is lucky, be comfortable with globbing. He probably does not understand regular expressions, but generally regular expressions allow a knowledgeable UNIX user a tremendous amount of power. Should applications omit regex searching, since the menu option for it might be confusing? If the goal is to recruit Joe User, the answer is probably yes. If the goal is to make the best possible system for techies, the answer is no.
It's not that it isn't a problem, it's just that the title of the paper implied that this was an Open-Source-specific set of problems. It's like saying in a car review "This car model has a significant problem -- it broke down after 250,000 miles. This lack of reliability has caused me a good deal of inconvenience." While this may well be true, the fact that *most* cars are likely to break down at or before 250,000 miles is a reasonable thing to point out. People don't want the criticism to be taken out of context, and to make a bogus decision based on that criticism.
Not only is it less glamorous, it's less fun and users complain about UI. I followed the gtk-gnutella mailing list when the UI was redesigned, and I remember how much complaining the UI guy got with each change he made.
This is not the case when you hit some obscure RPM bug.
What bug did you hit in RPM?
nope. It's UI-centric.
What?
I'm not sure I understand what you mean.
I would say that Office is indeed feature-centric. As a matter of fact, the feature-centric argument is the one argument that I don't agree with at *all* in the article. Commercial software needs bullet points to put on the box. Refactoring source code is hard to justify. I see people cleaning up code *all the time* in the open source world.
_THAT_ is the authors point exactly.
I dunno. I think the "design for self" issue is a valid complaint, but that it isn't generally *that* awful in practice. For instance, Samba isn't all that user-friendly to someone used to a GUI Windows environment -- the authors designed to to be efficient and powerful, but weren't worried about trying to bundle a GUI. It just means that other people make front ends.
I still cannot get my network printers working properly with CUPs.
That's really, really funny. I heard the bitter complaining for ages about CUPS -- how hard it is to configure, how many problems it causes. Yesterday, *I* found myself finally needed to set up printing on a box that had CUPS on it rather than LPRng or something similar.
I ran system-configure-printer. A GUI interface came up that asked me what I wanted to call the print queue. Well, I guess that I admit that I know that it's okay to call a print queue whatever I like from UNIX days, but I think that the dialog wasn't too confusing about making up a name, and listed all the constraints on me. It let me enter an optional description.
On the next screen, I finally managed to figure out why the CUPS configuration seems to confuse so many people. When you click "help", instead of giving you "help" on the entire dialog as I had expected, help for just the currently-selected menu option comes up -- LPD, CUPS, locally-connected, etc. That was unexpected on my part. However, I happend to choose the one I wanted to know about before hitting help, and so didn't have any problems (until I went back to examine the interface for UI issues just now). I remembered ESR getting CUPS and LPD confused, so I hit "help" when I happened to have CUPS up, and realized that I wanted LPD. I chose LPD.
I was given the option to enter an IP and queue name. I know the IP of the printer, and I don't think that it's all that exceptional to need to enter an IP -- any system being set up these days is going to need at least a *couple* of numbers or some basic configuration done when being set up. The queue name threw me for a minute, so I clicked "help" and was informed that the right option is usually "lp". I guess it would have been nice to have the default value filled in, but I don't think anything on this screen was *that* hard to manage.
I was then asked to select the printer type. The documentation said that "many people may have to select Generic Postscript", making it sound like a sub-par option, when it was, in fact, exactly what I wanted. I guess that was a bit confusing. It might be better, since this is just a high-level GUI config utility, to try to maintain a database of vendors and models, and choose the appropriate setting, or perhaps have an "unlisted" entry.
Finally, the thing told me that it was going to print a test page. I clicked OK, and a test page spat out of the printer. Seemed to work fine to me. I can't see what's so hard about CUPS, really. I've had *much* more complex problems trying to convince Windows to do something (and, admittedly, with other Linux software) than with CUPS.
The big advantage of open soure - sharing and code reuse - just isn't happening. For example, I'm looking for a new OSS webmail program. I have certain criteria. There are something like 209 projects related to this on Freshmeat. Why? That is no more efficient than a couple hundred closed-source vendors doing separate implementations of competing products. What have we gained?
Just out of curiosity, were you able to find a webmail package that you liked?
Is this an open-source-specific issue? It seems to me that there were a tremendous number of webmail packages, both open source and closed source, produced during the Internet boom. The difference is that a lot of the proprietary webmail vendors have folded, but the source to the open source projects still is available for anyone who would like to adopt.
Honestly, I don't think Freshmeat is the greatest system in the world for recommending a good "currently-best-available" software package. It really isn't intended to do so -- it's an attempt to be a complete catalog. I will grant that the Open Source world could really use a "this is what I currently consider best to do X, and this would be my second choice" website.
2. Choice really isn't, sometimes. When I am faced with two OSS projects that supposedly perform the same task, I often find that each lacks a function that the other has, and neither do everything they should. Why don't they team up and make a single, solid project?
For the same reason that all countries don't get together and make one world government. Robustness is important, and it's generally best gotten by having a number of people taking their own approach. Suppose one team is dead set on C++, and another thinks that servers are better written in Ocaml -- there's a pretty fundamental divide there, which can only be resolved by the creation of two projects. The Ocaml team thinks that the use of C++ will lead to too many security flaws, and the C++ team thinks that the use of Ocaml will lead to not enough coders and hence not enough functionality. Either one might be right -- the only way to find out is to start implementing. I'm glad that these different approaches exist -- without this, we'd all be out a lot.
Also, one of the tasks of a maintainer is when to say that a feature should go and when to say that a feature should say. This role is crucial for any kind of disciplined development, and is necessary even in the more relaxed open source world. Suppose you *really* want a feature in KDE but the KDE maintainer tells you that your idea is bad, and that you should piss off. You can then add your idea to GNOME, and see what happens. If your idea works out, eventually it's likely that KDE will adopt it as well.
The more Windows-like Linux becomes, the less I am interested in running it. I've mostly switched to FreeBSD (I used to be Redhat only).
I use Fedora, and I'm curious as to what elements of RH bother you.
I use sawfish, xbindkeys, and gkrellm for a desktop. I use xemacs and a bit of vi. Sure, the default environment is GNOME, but it's not as if you're constrained to use just GNOME.
I do agree that there are some things that irritate me that have been done in the name of ease of use.
For example, GTK/GNOME 2 both have user-rebindable accelerators disabled by default, and require config file modifications to re-enable them. This was done strictly to assist newbie users, and represents the removal of extremely powerful functionality that advanced users must hunt around to re-enable.
I've had cell phones for about 6 years now and they have all stayed off, why? because I'm more important than any other jerk that wants to ring me, there is hardly any good reason to use them, if somebody wants to reach me email is the preferred method, I'll check it several times a day and I don't have to listen to their endless babble.
I used to think roughly that (with the landline) and just didn't answer the phone, and referred people to email, checking the answering machine once a week or so. That was before I decided that *that* was too generous. I never read my email more than once a day, and after it kicked up in volume I started handling it about once a week to once every couple of weeks. The people that *really* know me can reach me via instant messaging protocols when I'm in front of my computer, and my employer can reach me at work (as can people with work-related communications). Other than that, I'm not a butler, and I don't see why masses of people all around the world should be able to ring me on demand (especially advertisers). My time is my time -- not time to be spent at the beck and call of others.
Plus, if people find that they can't immediately contact you, they have a tendancy to try to resolve problems themselves -- normally it's just "easier" to ask someone else every five seconds, which ends up overall wasting time (from communication overhead and context switching overhead).
Really? What makes you think OSDN won't start charging for SourceForge.net in the future? If so, what's your plan to migrate your projects to another server? If the SourceForge code were available, anybody would have the Freedom to start up a successor. That's what GForge is for.
Don't you view that as a sort of paranoid approach?
I mean, sure, if GForge works as well as SF, then it might be a good choice. But slamming SF because you think that they're suddenly going to clamp down on all the data they serve, without any evidence to support such a claim...that's paranoid. I recognize that there is a risk, but there's risks everywhere -- it's essentially impossible to eliminate everything. What if the FSF goes bad? Perhaps more plausibly, what if Stallman has a heart attack and a corporation manages to gain control (via bribery or whatnot) of the FSF, and hence has the freedom to write the GPLv3, with which GPL software by default can be automatically moved to. There was some discussion of giving special privileges to GPL-friendly companies, which IIRC what started Linus on his "I'm releasing my software under GPLv2 only" kick. The GPL controls a phenomenal amount of IP, probably more than any other legal document in history -- the ability to affect that document would be worth almost anything.
We have no way of knowing for certain who will turn out to be a baddie. We just try to generally reduce risk and react as things happen. Thus far, I've been pleased with the services provided by SourceForge, and have no reason to think that they will change their stance in the future.
The only difference between nongnu.org projects and gnu.org projects is that Free Software Foundation Inc owns the copyright in the source code of gnu.org projects.
Sure, but given all the domain names in the world, Savanah chose to use the domain name "nongnu.org", and proposes that software projects on SourceForge move there. Frankly, that's a political message that they propose that an awful lot of people should look at each day.
For example, GCC is under constant pressure by RMS to move from its own server (that happens to be hosted at Red Hat) and onto Savannah.
I really love it when people criticise Red Hat as not being oriented enough toward the community, and ignore the constant hosting, patches, technical work, donations, and political oomph and Red Hat provides to the open source world.
I wouldn't work on an Open Source project that required me to pay to work on it. It's just not reasonable.
I can understand them providing additional services, like POP3 email access @sourceforge rather than just email forwarding, or something like that, for money. However, if SF tries doing something like this, they are, simply and plainly, going to go away.
Savannah is lesser used -- there are fewer adherents of Free Software than Open Source.
The Open Source stance (as exemplified by ESR) is a more pragmatic one than an ideological one -- that people should use Open Source rather than Free Software because it *works better* than closed source, not because of a moral or philosophical mandate. The primary issue that SourceForge detractors bring up is that the current codebase is not available; this is an issue to a number of people strongly ideologically aligned with Free software, who want to interact with nothing but Free software. There is a parallel here. Since SF costs nothing, works well, and helps spread and facilitate open source software, there are few pragmatic issues with SourceForge that Savannah solves. Thus, the issues with Open Source that Free advocates have are mostly the same complaints that are raised about SourceForge.
Savannah's main issues are caused by a lack of people working on it, and it is currently less ready-to-go than SourceForge. It's HURD and Linux in a mirror.
Savannah makes its feelings on the importance of Free software very clear with the nongnu and gnu names. The SF people don't particularly place a lot of emphasis on someone being associated with a project or having a particular license -- there's no sourceforge.sortaopen.net for BSD-licensed projects, for instance.
Finally, while this is more germane to this story than to SF in general, the politics in the linked-to story remind me a good deal of the complex and never-ending debates about Free software purity that come up more frequently in the Free Software world.
I suppose that a lot of Free advocates are going to view this as a bit flamish -- I guess it's a bit cutting in that it identifies that Savannah hasn't been operating as well as SourceForge, but I don't feel that it's particularly false or misleading.
I use the GNU utilities as well as Apache every day -- I like both chunks of software.
I also, as people who read my posts frequently know, tend to often feel a bit frusterated with Free advocates. I do, not infrequently, think that Free folks can come off as a bit too rabid to the general public -- this mainly becomes an issue when media, desperate for some kind of figurehead for the open source world, settle on RMS, and he propagates his (intimidating to a CTO) views on intellectual property. I also remember when the Crystal Space team (an excellent LGPLed 3d engine), wanted to be absolutely correct WRT the GPL and valuing Stallman's input, wrote him to ask for a bit of clarification on a licensing detail. Stallman's response, an enlightening read, highlights a good deal of what I consider the difference between Open Source folks like Jorrit and Free folks like Stallman.
For one thing, people may not realise just how annoying they're being; this lets them know, and gives them a chance to improve their manners.
If they are willing to go out of their way to be polite. The current solution *is* to ask people to behave differently, and it's clearly not working. Oh, it's fine for the rude users, but it works much less well for those people who have to put up with it.
Furthermore, it requires continual effort on the part of the people being imposed upon -- having to go up and hassle with some stranger on the subway each day is not anyone's idea of a good morning.
A jammer may have be shiny and give you a feeling of power, but it's still a bad solution to the problem.
"shiny and a feeling of power, but a bad solution to the problem it tries to solve"...that description is terribly ironic, as I would have used almost the same criticism, word for word, about cell phones and useful communication.
AFAIC, a loud and annoying mobile phone ring is no better or worse than an equally loud and annoying digital watch tune, and a loud and annoying conversation on a mobile phone is no better or worse than an equally loud and annoying conversation of any other kind.
Yes, but:
(a) most people don't use digital watch alarms constantly -- many bored people call other people to socialize all day long.
(b) digital watch alarms aren't as loud, and are simple and easy to tune out.
(c) the conversation, as the story pointed out, is much worse than the ring tone.
I just happen to feel that generally, technical problems are better than social ones, and require everyone involved to do less work.
Instead of expecting every person being annoying to have to try to convince an irritating cell phone user to stop, the phone conversation is simply jammed. This shifts the onus of putting out effort from the people who don't want to be annoyed by phones onto phone users to be less annoying and companies to sell less annoying products. You don't want to be jammed, speak quietly, politely, and use a vibrator.
Just as I don't think it's feasible for cell phone opponents to fight technology -- cell phones are going to be here and increasing in use, and people are going to be assholes with them, I don't think it's feasible for cell phone advocates to fight technology -- a jammer need not be very large or powerful, and is essentially impossible to effectively track down.
I consider SourceForge to be representative of Open Source Software, and Savannah to be representative of Free Software.
It's amazing how accurately they seem to portray their respective ideologies.
Yes, it's like 1024 chickens pulling 1024 little plows.
But if the 1024 chickens do it at a lower cost and at a faster speed, why would he bother with 2 oxen?
I think you didn't read his post in entirety.
This is the equivalent of posting, "My AMD processor is better than your Intel processor!" It's a quote designed to ignite a fact-less argument on who has bigger ones.
No, no. That *would* have content -- it would mean that some consumer, for one reason or another, is willing to back a purchase he made. *This* is contentless -- an exec at a company says that really, his products (which are being squashed in the marketplace) are better than the competition. You'd expect that from any company exec.
You have to be impressed that they hold 19th, really. I thought that Cray was long since out of business.