### I'm disappointed by Google's use of the 'goto' keyword.
While goto is often better avoided, a call like "goto error;" is among the perfectly valid uses of goto, since it actually can make code more clear and logical then code without goto. Such use of goto is really no different then exceptions in C++, simply a way to get to the place that handles the error conditions without having to painfully drag error-variables through the code.
I seriously doubt that. Nostalgia might be the thing that gets you started playing older games or getting interested, but its not the thing that keeps you playing for hours and hours, gameplay is the thing that does. I didn't play XCom:UFO a year ago for like 40h because I had some found memories of the past about it, no, I had never ever touched or seen that game before. I played it because it simply is a great game. The graphics might be a bit dated, but the stuff you can do in that game is well beyond must current games. Same with DeusEx or Cybernator. Also recently played Metroid(NES), even so I still hadn't finished MetroidPrime2(GCN), why? Because the original is a much more fun game for me, less frustration and more challange and takes far less time to get into. Same thing is true for basically all Mario games (except Sunshine), easy to get into, fun to play. This doesn't mean that all games of the past are great, I did play HalfLife, which I hadn't played before, it however left me rather unimpressed, nice AI and the one-huge-level thing might been good for its time, but I found the whole story rather dull and gameplay rather repetative.
To make it short, what makes the old games for me interesting is that they are simply more direct then todays games. In old 2D jump'n runs there is often only one direction you can walk into, and that is forward, you don't have to hunt for half an hour just to find the entrance for a level like in some of todays games, meaning you can switch them on, play for a 10mins and have fun, something not possible with many todays games.
Another thing that I also find interesting about past games is that they are, what I would call, 'multi-mode', meaning they have multiple levels of gameplay. In Xcom:Ufo for example you have the round based stratagy part, but you also have the part where you have to intercept Ufos and you have to take care of your money, do research on weapons and build your base, which makes it for me much more interesting then something like FF:Tactics.
And there are of course also the games that focused on story or athmosphere, Indy4, MonkeyIsland, AnotherWorld and the like that are still pretty much unmatched, even today. Most games today simply do story as some side-thing and fill the 10h of gameplay with repetative and boring gameplay.
Doesn't mean that all todays games are bad, Ico, Beyond&GoodEvil, PrinceOfPersiaSoT1, OperationFlashpoint are all great games on a similar level like some of the old classics, but todays 'blockbuster' games leave me all rather unimpressed.
In the end its probally that we only remember the great games of the past and not the bad ones, so you compare the great games of the past with the average game of the present. However the games of the past are often still great even today. If you can live with 320x200 graphics many of them still provide a gameplay experince that is as good if not better than the best games you can find today.
### I would also prefer if the Xbox and GC controller featured parallel analog sticks, like the PS2 controller.
Please not, those are probally the worst thing invented in controller history. The thumb twisting required to reach them is just not healthy. I don't mind that much for a stick that I don't have to often, like the C-stick, but having the primary analog stick in that position is causing nothing but pain if I play for longer periods of time, heck even with short periods of play I find that position nothing but unconfortable.
In times where basically all games are played with analog stick the stick belongs into the primary position where it is easy to reach, not at the edge of the controller and I am rather suprised that Sony choses the same bad position again for the PS3.
What I would find most interesting would be an upgrade in the framerate, some of the early 3D titles on the SNES and later the N64 were just painfully slow and suffered from huge drops in the framerate. I would love to be able to replay StarFox or StuntRaceFX with constant 60fps.
Its true that Mario2 didn't started out as a Mario game, but lots of ideas, enemies and concept that where first found in Mario2 have been carried over into other Mario games (Shyguys in Yoshi, object lifting in Mario vs DonkeyKong, different character abilities in Mario64DS, etc), so that Mario2 is now actually very well integrated into the total series of Mario games. Not calling it a Mario game kind of dismisses the huge impact it had on the series as a whole.
### The reason you shouldn't switch off your computer is to keep the electronics at a relatively constant temperature.
I seriously doubt that this is an issue. I have yet to actually see a single computer that breaks for this reason. Fans, harddisk and the like all break years before your electronics go by by. And a fan constantly rotating 24/7 for sure gets more used then one that only rotates for 40h a week.
### It doesn't have anything to do with constant access
Its *all* about constant access. If computers would be available seconds after you touched the power button, like C64 used to be, people wouldn't bother to let their computer running, noise, heat, powerconsumtion all are very good reason to switch the PC of as soon as you can, but with those long boot times we have these days its just to annoying to wait a minute or longer just to have a quick look at a webpage.
I wouldn't say so, "Batman Begins" seems to be quite good, some of "Animatrix" was great, "Escape From Butcher Bay" seems to be pretty good, watching the first few minutes of Indy3 was fun and Buffy/Angel also had a bunch of going back in time over the whole series that is pretty good as well and there is also "Knights of the old Repulic". Ok, some of this are actually games, not movies, but it shows that one can do plenty of interesting stuff with story happening before some movie.
The problem with the StarWars prequels was that they simply hadn't much story of their own, boring CloneWars here, stupid Jedi there and a bunch of pointless back-references all over the place (look that ship looks almost like a tie-fighter and that almost like a x-wing and there is Chewbacca) and worst of all they simply missed out many interesting parts. Grievous for example seems to be a rather interesting character, in the movies however we don't get to know much about his background at all. The CloneWars cartoons actually show some of it and it would have been very nice to see some of that on the big screen. And after all the prequels simply fail the whole reason why they where created in the first place, to show us why Anakin got Darth Vader, this most important point was simply rushes over in a few minutes (Imperator: wanna be evil? Anakin: Yep).
### Which brings to mind.. how do you observe something without changing it?
Easy, use a big-fat telescope and just look into the sky, voila, the past, whole sky is full of it, since light travels at a finite velocity everything you observe there is already long gone the moment its light arive here at earth. Doesn't really have much todo with time travel, but it shows how you can observe without a way to change it, simply requires that it already was observed and the information what was observed is still floating around there somewhere.
Actually the best you can do is go for *both* MS Natural and Dvorak, if all the other keyboards you use are normal non-split keyboards you will have basically very little or no problems at all switching keyboards. The reason for that is that the brain kind of remembers the hand position in combination with the keyboard layout, so if you sit infront of a split-keyboard the brain will switch to Dvorak mode and infront of a non-split it will go to Qwerty mode. Learning Dvorak doesn't magically erase all your Qwerty knowledge, especially if you still use Qwerty regular, its only a matter to have the environments for both layouts reasonably different (split vs non-split, laptop keyboard, etc.) to make the switching easy.
LSB is only about binary compability and doesn't even get that nicely solved for political reasons (hint: lsb-rpm). Beside from that LSB is really only about writing that down that already almost everybody does, they don't really 'create' anything new or innovate.
The one group that really does move stuff at the moment is freedesktop.org and while they are slow at times, progress is steady.
### One of the most important tricks of using spatial filemanagers is to use the very feature that makes the thing spatial: set suitable view options, icon placement, emblems, background, window position and window size for the folders (or at least for the most important ones)
And that is exactly the thing that can't work with todays filesystems. When I browse to/usr/local/share/doc/something-someversion I will browse there once, maybe twice in my lifetime, I won't *ever* visit that directory frequently. So I will never ever set it up nicly, remember where the spatial window poped up or anything. All those settings will seem to be completly *random* the next time I visit that place, which will most likly be weeks or month after the last time. And with huge diretory trees there are lots and lots of places that I won't visit for month or ever. So if I ever try to browse there I get randomly poping up windows with random settings allocated over the years, but completly useless to me at the current point in time.
Spatial works only good if the underlying directory structure is tuned for it, as it was on the Amiga, Unix is basically as far away from that as you can get, making spatial nothing but pain.
### I have to say boy that old dialog is clunky and annoying.
Heck, yes, the old dialog is clunky and annoying, no doubt about that, everybody will agree with you. It however provided one feature that worked and worked great and that was Tab-completion. And this one feature got pretty much broken and dumped down in the new one. That Ctrl-L hack in the new one makes the old one really look like a beauty.
## I basically never write filenames to the dialog. In my opinion it sort of like defeats the whole puprpose of having graphical UI.
Well, you give up the by far most effective way to browse deep directory structures by ignoring textual input and tab-completion. Don't get me wrong I, really like good mouse driven interfaces, but what I really can't stand are interfaces that render the keyboard unusable for actually no good reason at all.
### Joe the Average opens preferences dialog,
Come on, that argument is now really nothing but bogus. Linux doesn't *HAVE* Joe Average as a user, but tons of people that will have zero problem with a new option deep down in the preferences. Gnome people should take a little bit more care about their actual userbase and don't write for some imaginary totally dumb user.
I for one still look back at the golden days where of Gnome1.4, where lots of things might have been a bit ugly, but where software was actually written for those people that use it.
### Linux users *assume* that ALL spatial interfaces are equally botched.
Well, they know that the Gnome way of spatial browsing is rather unusable, thats all what this is about. They aren't bitching about some theoretical thing, but that file browser that is annoying them on their desktop right now today. If there is some true magically way to make spatial browsing fun, then the Gnome people should implement it and show it to the world, fact of life is however that the current implementation is nothing but a PITA.
In the end I simply doubt that spatial browsing is of much use today, when you have few files and few folders its great, but I fail to see how you can browse deep directory structure with it without cluttering your desktop with zillions of windows. And no, Ctrl-Clicks doesn't really help much, sinc e then you are left with windows poping up at random locations which is really not fun, I prefer to that the Rox way of browsing any day.
Even bittorrent needs the 'fat piepes', it can't magically create bandwidth out of nowhere. Only thing that bittorrent helps with it that those that publish the file in the first place don't need a 'fatpiepe', but those who download can provide them. If you want to distribute to the masses you simply need a bunch of people that can upload more then they download, if you are stuck with ADSL people who can download at 100kb/s or more, but only upload at 15kb/s at best you naturally run into problems, even with bittorrent.
### Some distro, company, or group is gonna have to take Linux and do what Apple did for bsd, except leave the contributions open.
Not really, at least not in the way current distros do it. Today you get lots of nice stuff, automatic hardware detection, tools to make configuration easy and such with each and every distro, problem is that they all write their own tools, which are basically hacks ontop of the cludge that is already there. That however causes tons of problems, config files that get overwritten by those tools, howtos that just don't work on your distro, incompatibilities all over the place.
What we wood need is fixes at the base, consistent config file syntax, consistent naming of files, a cross distro packaging system and such. Since that is what is really annoying and makes writing proper gui configurators hard.
The problem with spatial is that it only works good with very few folders, meaning it worked great in the days of Amigas where you only had floppies to store your stuff, today however where people have multiple 160GB drivers and actually fill them, spatial browsing just can't work much good, essentially its useless for browsing deep directory trees. It had nothing todo with bitching, its a simple fact of life that for many people spatial browsing just can't work, ever.
The problem is also not that the 'blowhards' rejected the paradigm, but the problem was that the gnome developers didn't even provide a easy way to switch it off, but only obscure registry hacks. They did the same with the new file dialog, yes, the old one really needed work, lots of work, the new one however threw the one feature away that made the old one great, tab-completion and only only got it back in a sucky, almost unusable, cludge of an implementation. I don't doubt that in a few gnome releases the file dialog will be great, but the problem is that the Gnome developers constantly piss of their userbase and only fix it month later.
There is nothing wrong with providing new features, but they should really try a little bit harder to make the upgrade path a little bit more smooth and get rid of their "we know better what is good for you then yourself".
### is it easier to select all the parameters (click here, type there) in a VB form || open a terminal and type 'wget..'???
Depends if you do it daily or only once every few weeks. Typing the parameters is one things, remembering them all a completly different issue. When I use wget, and I use it a lot, I still end up having to look up the more obscure option up in the man-page, in a GUI I could simply click on them. Thats why a good programm should provide both a interactive way and a batch way to access its functionality.
That Microsofts complete refusal in the past to provide a good shell has done far more harm then good is no doubt about that, but when I look at what the new MSH has to offer I have to say that the old rusty bash could need a few improvments as well. There is no reason why textual access to functions should be limited to lausy 80x24 terminals, I want a Emacs like 'M-x' command line as a standard part of every programm or even better a full implementations of the ideas of Jeff Raskins THE, which uses both keyboard and mouse for the benefit of the user.
### When I have to work on a repetitive task I can either write a macro or have the exact sequence of key-strokes down and do the job much faster.
Back then in 1995 I used to use CorelDraw a lot, when I wanted a macro I pressed the 'record'-button in the macro-dialog, did what I want to do and pressed stop. Once done I could rerun that thing as much as I like. If that wasn't enough, I simply could imported the macro into the included script editor and tweak it to my liking, change parameters, even add a GUI or whatever. That was by far the most comfortable way to generate scripts I have seen so far. Just because todays GUIs suck pretty much when it comes to automatisation doesn't mean it can't be done in a way that is perfectly accessible with only the mouse and at the same time gives you all the power that you have from a real scripting language.
The throuble is that todays userinterfaces either focus completly on the mouse or the keyboard, very few manage to use both to the advance of the user.
### And you just *can't* force all developers to use the same build system.
Certainly true, but if something like freedesktop.org would publish a recommendation I sure at least quite a few would follow. At the moment I couldn't follow any standard way even if I want to, because there simply isn't any.
### And you just *can't* force all developers to use the same build system.
Of course you can't force them to use a different build system, thats why I said 'metainformation', by that I mean a common way to document things like how to install a package, how to compile it, how to install it into a different prefix, destdir, which libraries it requires, etc. things you can do with basically every build system somehow, they are just named often quite different.
The source packages that you find for Debian or Redhat do basically exactly that, they wrap the underlying build system into something that is usable for an automatic build of the binary package. It would be nice if that part would be somehow become standard across distros so that I could just say 'build --prefix=/foo/', instead have manuallly trying to figure out if its 'make -f Makefile.linux PREFIX=/foo' or './configure --prefix=foo' or something completly different.
### Any time anyone tries to propose a "one size fits all" solution, I just have to shake my head. Not everyone agrees on what's the best way to do something. Not everybody *should*, otherwise no progress is made.
Certainly true, but the different build systems aren't really that different in *what* they do, they are only different in *how* they do it. And since the 'what' part (install, build, cleanup) is basically offered by all of them, some 'metainformation' could be used to allow to automate this process.
Starting from such 'metainformation'-enhanched source tarballs it would then be possible to automatically build binary packages for lots of architectures and such.
### If you let users install system software then we go back to the days of rampant viruses and worms...
Users already can install and run viruses and worms as much as they like on Linux. Lack of install support for users in apt-get isn't a security feature, since there are numerous other ways to install software. I would go even as far as saying that lack of user install in apt-get is a huge security risk in the long run, since it forces user to look for random other sources for binaries instead of using the trusted Debian repository.
### you'd know about adding APT respositories to your sources.list file
If you would have acutally used Debian with unofficial debs you would now that Debians dependency tree is pretty fragile, if install not-so-well-build unofficial debs you can have lots of fun cleanup up later on. Yes, it sometimes works well, but often does not.
### My systems are always up to date and only occasionally have problems
If you think running without security updates is a good thing than that is your problem.
### For people who want new0ish software but don't like the idea of running something "unstable" there's the testing branch.
People want *some* new software (say a new Gimp or KDE or whatever they need most for they daily work), they don't want to have a completly bleeding edge that doesn't even provide security updates.
### How are we supposed to have the time/energy/desire to build binaries for all these arches, much less hardware to test them on?
For the beginning standard source format would already help a lot, ie. one with computer readable metainformation on how to build, not just human readable INSTALL files. The rest could then be simply autobuilded.
### but there are good technical reasons why it can't be as simple as we'd like.
I wouldn't call it 'good technical reasons', most of it is simply historic and lack of standardization. Nothing would really stop a standard package format to work out on a large scale if people simply tried a little bit harder.
### You definately should not be able to install it system-wide no matter what package management system you have.
Of course you should not, unless you are root, thats now something bloody obvious that I didn't assume that it would be needed to mentioned specially. What I was talking about is installing stuff in the users home directory.
### If you want a cutting edge version of gimp, download and compile it in your home directory.
Why should the user be forced to compile and track dependency manually when the Debian repository provides a perfectly working version of it? Why can't apt-get allow him to install it into his home? Weren't package managers invented to avoid compiling from source in the first place?
### Sure debian stable is a little behind but remember, it's *stable* meaning that buggy cutting edge program won't be there.
You don't seem to have used Debian for too long, 'stable' refers to the package dependencies *not* to the software. There is plenty of software in stable that is far more 'bleeding edge' than unstable, simply because 'stable' was frozen at a time where the software just has jumped to a new major version and never got any of the bug fixes that the software in unstable got. After all only security fixes get down to stable, rest stays in testing and unstable. Anyway, thats a different issue.
### but I don't expect the author of a program to package their program up for each distro.
Neither do I, what I want is a way to package programms independly of the distro. Which either means source tar.gz + some computer readable metadata (as oposed to human readable INSTALL files like its handled at the moment) or autopackage/lsb-rpm style binaries.
### Why do you think that it's a sensible idea for users to be installing software in the first place?
Do you think its a better idea to run as root for every little piece of software you want to try out? What makes you think that its actually not a sensible idea to install software as user? Its a perfectly fine thing todo and its the way Windows and MacOS work for decades. You won't stop viruses from spreading by forcing users to either 'sudo' or use untrusted sources instead of official Debian repositories.
### In which case you can simply build the software to work in that environment.
But weren't package managers invented to avoid that in the first place?
Now thats a great argument. If that is really the case I would call that broken by design, but could you care to give some details why exactly it wasn't supposed to handle such a situation? After the only difference between installing as root and installing as user is the $PREFIX, something reasonably easy to fix if one tries to. There are of course packages that have to be installed as root (kernel and such), but they are by far the minority.
### You can install software in your home directory in Linux.
Yes, you can. However the distros packagemanager and the distros package repository is completly useless for doing so and thats exactly the point of why apt-get isn't a very good solution for handling software. It works great for some uses, but completely fails at others.
Slashdot Mirror
### I'm disappointed by Google's use of the 'goto' keyword.
While goto is often better avoided, a call like "goto error;" is among the perfectly valid uses of goto, since it actually can make code more clear and logical then code without goto. Such use of goto is really no different then exceptions in C++, simply a way to get to the place that handles the error conditions without having to painfully drag error-variables through the code.
I seriously doubt that. Nostalgia might be the thing that gets you started playing older games or getting interested, but its not the thing that keeps you playing for hours and hours, gameplay is the thing that does. I didn't play XCom:UFO a year ago for like 40h because I had some found memories of the past about it, no, I had never ever touched or seen that game before. I played it because it simply is a great game. The graphics might be a bit dated, but the stuff you can do in that game is well beyond must current games. Same with DeusEx or Cybernator. Also recently played Metroid(NES), even so I still hadn't finished MetroidPrime2(GCN), why? Because the original is a much more fun game for me, less frustration and more challange and takes far less time to get into. Same thing is true for basically all Mario games (except Sunshine), easy to get into, fun to play. This doesn't mean that all games of the past are great, I did play HalfLife, which I hadn't played before, it however left me rather unimpressed, nice AI and the one-huge-level thing might been good for its time, but I found the whole story rather dull and gameplay rather repetative.
To make it short, what makes the old games for me interesting is that they are simply more direct then todays games. In old 2D jump'n runs there is often only one direction you can walk into, and that is forward, you don't have to hunt for half an hour just to find the entrance for a level like in some of todays games, meaning you can switch them on, play for a 10mins and have fun, something not possible with many todays games.
Another thing that I also find interesting about past games is that they are, what I would call, 'multi-mode', meaning they have multiple levels of gameplay. In Xcom:Ufo for example you have the round based stratagy part, but you also have the part where you have to intercept Ufos and you have to take care of your money, do research on weapons and build your base, which makes it for me much more interesting then something like FF:Tactics.
And there are of course also the games that focused on story or athmosphere, Indy4, MonkeyIsland, AnotherWorld and the like that are still pretty much unmatched, even today. Most games today simply do story as some side-thing and fill the 10h of gameplay with repetative and boring gameplay.
Doesn't mean that all todays games are bad, Ico, Beyond&GoodEvil, PrinceOfPersiaSoT1, OperationFlashpoint are all great games on a similar level like some of the old classics, but todays 'blockbuster' games leave me all rather unimpressed.
In the end its probally that we only remember the great games of the past and not the bad ones, so you compare the great games of the past with the average game of the present. However the games of the past are often still great even today. If you can live with 320x200 graphics many of them still provide a gameplay experince that is as good if not better than the best games you can find today.
### I would also prefer if the Xbox and GC controller featured parallel analog sticks, like the PS2 controller.
Please not, those are probally the worst thing invented in controller history. The thumb twisting required to reach them is just not healthy. I don't mind that much for a stick that I don't have to often, like the C-stick, but having the primary analog stick in that position is causing nothing but pain if I play for longer periods of time, heck even with short periods of play I find that position nothing but unconfortable.
In times where basically all games are played with analog stick the stick belongs into the primary position where it is easy to reach, not at the edge of the controller and I am rather suprised that Sony choses the same bad position again for the PS3.
What I would find most interesting would be an upgrade in the framerate, some of the early 3D titles on the SNES and later the N64 were just painfully slow and suffered from huge drops in the framerate. I would love to be able to replay StarFox or StuntRaceFX with constant 60fps.
### Ok, but does bash or ksh run on windows?
Sure they do, like almost all Unix software, either via the full blown Cygwin or via the more lightweight MSYS.
Its true that Mario2 didn't started out as a Mario game, but lots of ideas, enemies and concept that where first found in Mario2 have been carried over into other Mario games (Shyguys in Yoshi, object lifting in Mario vs DonkeyKong, different character abilities in Mario64DS, etc), so that Mario2 is now actually very well integrated into the total series of Mario games. Not calling it a Mario game kind of dismisses the huge impact it had on the series as a whole.
### The reason you shouldn't switch off your computer is to keep the electronics at a relatively constant temperature.
I seriously doubt that this is an issue. I have yet to actually see a single computer that breaks for this reason. Fans, harddisk and the like all break years before your electronics go by by. And a fan constantly rotating 24/7 for sure gets more used then one that only rotates for 40h a week.
### It doesn't have anything to do with constant access
Its *all* about constant access. If computers would be available seconds after you touched the power button, like C64 used to be, people wouldn't bother to let their computer running, noise, heat, powerconsumtion all are very good reason to switch the PC of as soon as you can, but with those long boot times we have these days its just to annoying to wait a minute or longer just to have a quick look at a webpage.
### Is the whole concept of "prequels" flawed?
I wouldn't say so, "Batman Begins" seems to be quite good, some of "Animatrix" was great, "Escape From Butcher Bay" seems to be pretty good, watching the first few minutes of Indy3 was fun and Buffy/Angel also had a bunch of going back in time over the whole series that is pretty good as well and there is also "Knights of the old Repulic". Ok, some of this are actually games, not movies, but it shows that one can do plenty of interesting stuff with story happening before some movie.
The problem with the StarWars prequels was that they simply hadn't much story of their own, boring CloneWars here, stupid Jedi there and a bunch of pointless back-references all over the place (look that ship looks almost like a tie-fighter and that almost like a x-wing and there is Chewbacca) and worst of all they simply missed out many interesting parts. Grievous for example seems to be a rather interesting character, in the movies however we don't get to know much about his background at all. The CloneWars cartoons actually show some of it and it would have been very nice to see some of that on the big screen. And after all the prequels simply fail the whole reason why they where created in the first place, to show us why Anakin got Darth Vader, this most important point was simply rushes over in a few minutes (Imperator: wanna be evil? Anakin: Yep).
### Which brings to mind.. how do you observe something without changing it?
Easy, use a big-fat telescope and just look into the sky, voila, the past, whole sky is full of it, since light travels at a finite velocity everything you observe there is already long gone the moment its light arive here at earth. Doesn't really have much todo with time travel, but it shows how you can observe without a way to change it, simply requires that it already was observed and the information what was observed is still floating around there somewhere.
Actually the best you can do is go for *both* MS Natural and Dvorak, if all the other keyboards you use are normal non-split keyboards you will have basically very little or no problems at all switching keyboards. The reason for that is that the brain kind of remembers the hand position in combination with the keyboard layout, so if you sit infront of a split-keyboard the brain will switch to Dvorak mode and infront of a non-split it will go to Qwerty mode. Learning Dvorak doesn't magically erase all your Qwerty knowledge, especially if you still use Qwerty regular, its only a matter to have the environments for both layouts reasonably different (split vs non-split, laptop keyboard, etc.) to make the switching easy.
LSB is only about binary compability and doesn't even get that nicely solved for political reasons (hint: lsb-rpm). Beside from that LSB is really only about writing that down that already almost everybody does, they don't really 'create' anything new or innovate.
The one group that really does move stuff at the moment is freedesktop.org and while they are slow at times, progress is steady.
### One of the most important tricks of using spatial filemanagers is to use the very feature that makes the thing spatial: set suitable view options, icon placement, emblems, background, window position and window size for the folders (or at least for the most important ones)
/usr/local/share/doc/something-someversion I will browse there once, maybe twice in my lifetime, I won't *ever* visit that directory frequently. So I will never ever set it up nicly, remember where the spatial window poped up or anything. All those settings will seem to be completly *random* the next time I visit that place, which will most likly be weeks or month after the last time. And with huge diretory trees there are lots and lots of places that I won't visit for month or ever. So if I ever try to browse there I get randomly poping up windows with random settings allocated over the years, but completly useless to me at the current point in time.
And that is exactly the thing that can't work with todays filesystems. When I browse to
Spatial works only good if the underlying directory structure is tuned for it, as it was on the Amiga, Unix is basically as far away from that as you can get, making spatial nothing but pain.
### I have to say boy that old dialog is clunky and annoying.
Heck, yes, the old dialog is clunky and annoying, no doubt about that, everybody will agree with you. It however provided one feature that worked and worked great and that was Tab-completion. And this one feature got pretty much broken and dumped down in the new one. That Ctrl-L hack in the new one makes the old one really look like a beauty.
## I basically never write filenames to the dialog. In my opinion it sort of like defeats the whole puprpose of having graphical UI.
Well, you give up the by far most effective way to browse deep directory structures by ignoring textual input and tab-completion. Don't get me wrong I, really like good mouse driven interfaces, but what I really can't stand are interfaces that render the keyboard unusable for actually no good reason at all.
### Joe the Average opens preferences dialog,
Come on, that argument is now really nothing but bogus. Linux doesn't *HAVE* Joe Average as a user, but tons of people that will have zero problem with a new option deep down in the preferences. Gnome people should take a little bit more care about their actual userbase and don't write for some imaginary totally dumb user.
I for one still look back at the golden days where of Gnome1.4, where lots of things might have been a bit ugly, but where software was actually written for those people that use it.
### Linux users *assume* that ALL spatial interfaces are equally botched.
Well, they know that the Gnome way of spatial browsing is rather unusable, thats all what this is about. They aren't bitching about some theoretical thing, but that file browser that is annoying them on their desktop right now today. If there is some true magically way to make spatial browsing fun, then the Gnome people should implement it and show it to the world, fact of life is however that the current implementation is nothing but a PITA.
In the end I simply doubt that spatial browsing is of much use today, when you have few files and few folders its great, but I fail to see how you can browse deep directory structure with it without cluttering your desktop with zillions of windows. And no, Ctrl-Clicks doesn't really help much, sinc e then you are left with windows poping up at random locations which is really not fun, I prefer to that the Rox way of browsing any day.
Even bittorrent needs the 'fat piepes', it can't magically create bandwidth out of nowhere. Only thing that bittorrent helps with it that those that publish the file in the first place don't need a 'fatpiepe', but those who download can provide them. If you want to distribute to the masses you simply need a bunch of people that can upload more then they download, if you are stuck with ADSL people who can download at 100kb/s or more, but only upload at 15kb/s at best you naturally run into problems, even with bittorrent.
### Some distro, company, or group is gonna have to take Linux and do what Apple did for bsd, except leave the contributions open.
Not really, at least not in the way current distros do it. Today you get lots of nice stuff, automatic hardware detection, tools to make configuration easy and such with each and every distro, problem is that they all write their own tools, which are basically hacks ontop of the cludge that is already there. That however causes tons of problems, config files that get overwritten by those tools, howtos that just don't work on your distro, incompatibilities all over the place.
What we wood need is fixes at the base, consistent config file syntax, consistent naming of files, a cross distro packaging system and such. Since that is what is really annoying and makes writing proper gui configurators hard.
The problem with spatial is that it only works good with very few folders, meaning it worked great in the days of Amigas where you only had floppies to store your stuff, today however where people have multiple 160GB drivers and actually fill them, spatial browsing just can't work much good, essentially its useless for browsing deep directory trees. It had nothing todo with bitching, its a simple fact of life that for many people spatial browsing just can't work, ever.
The problem is also not that the 'blowhards' rejected the paradigm, but the problem was that the gnome developers didn't even provide a easy way to switch it off, but only obscure registry hacks. They did the same with the new file dialog, yes, the old one really needed work, lots of work, the new one however threw the one feature away that made the old one great, tab-completion and only only got it back in a sucky, almost unusable, cludge of an implementation. I don't doubt that in a few gnome releases the file dialog will be great, but the problem is that the Gnome developers constantly piss of their userbase and only fix it month later.
There is nothing wrong with providing new features, but they should really try a little bit harder to make the upgrade path a little bit more smooth and get rid of their "we know better what is good for you then yourself".
### is it easier to select all the parameters (click here, type there) in a VB form || open a terminal and type 'wget ..'???
Depends if you do it daily or only once every few weeks. Typing the parameters is one things, remembering them all a completly different issue. When I use wget, and I use it a lot, I still end up having to look up the more obscure option up in the man-page, in a GUI I could simply click on them. Thats why a good programm should provide both a interactive way and a batch way to access its functionality.
That Microsofts complete refusal in the past to provide a good shell has done far more harm then good is no doubt about that, but when I look at what the new MSH has to offer I have to say that the old rusty bash could need a few improvments as well. There is no reason why textual access to functions should be limited to lausy 80x24 terminals, I want a Emacs like 'M-x' command line as a standard part of every programm or even better a full implementations of the ideas of Jeff Raskins THE, which uses both keyboard and mouse for the benefit of the user.
### When I have to work on a repetitive task I can either write a macro or have the exact sequence of key-strokes down and do the job much faster.
Back then in 1995 I used to use CorelDraw a lot, when I wanted a macro I pressed the 'record'-button in the macro-dialog, did what I want to do and pressed stop. Once done I could rerun that thing as much as I like. If that wasn't enough, I simply could imported the macro into the included script editor and tweak it to my liking, change parameters, even add a GUI or whatever. That was by far the most comfortable way to generate scripts I have seen so far. Just because todays GUIs suck pretty much when it comes to automatisation doesn't mean it can't be done in a way that is perfectly accessible with only the mouse and at the same time gives you all the power that you have from a real scripting language.
The throuble is that todays userinterfaces either focus completly on the mouse or the keyboard, very few manage to use both to the advance of the user.
### And you just *can't* force all developers to use the same build system.
Certainly true, but if something like freedesktop.org would publish a recommendation I sure at least quite a few would follow. At the moment I couldn't follow any standard way even if I want to, because there simply isn't any.
### And you just *can't* force all developers to use the same build system.
Of course you can't force them to use a different build system, thats why I said 'metainformation', by that I mean a common way to document things like how to install a package, how to compile it, how to install it into a different prefix, destdir, which libraries it requires, etc. things you can do with basically every build system somehow, they are just named often quite different.
The source packages that you find for Debian or Redhat do basically exactly that, they wrap the underlying build system into something that is usable for an automatic build of the binary package. It would be nice if that part would be somehow become standard across distros so that I could just say 'build --prefix=/foo/', instead have manuallly trying to figure out if its 'make -f Makefile.linux PREFIX=/foo' or './configure --prefix=foo' or something completly different.
### Any time anyone tries to propose a "one size fits all" solution, I just have to shake my head. Not everyone agrees on what's the best way to do something. Not everybody *should*, otherwise no progress is made.
Certainly true, but the different build systems aren't really that different in *what* they do, they are only different in *how* they do it. And since the 'what' part (install, build, cleanup) is basically offered by all of them, some 'metainformation' could be used to allow to automate this process.
Starting from such 'metainformation'-enhanched source tarballs it would then be possible to automatically build binary packages for lots of architectures and such.
### If you let users install system software then we go back to the days of rampant viruses and worms...
Users already can install and run viruses and worms as much as they like on Linux. Lack of install support for users in apt-get isn't a security feature, since there are numerous other ways to install software. I would go even as far as saying that lack of user install in apt-get is a huge security risk in the long run, since it forces user to look for random other sources for binaries instead of using the trusted Debian repository.
### you'd know about adding APT respositories to your sources.list file
If you would have acutally used Debian with unofficial debs you would now that Debians dependency tree is pretty fragile, if install not-so-well-build unofficial debs you can have lots of fun cleanup up later on. Yes, it sometimes works well, but often does not.
### My systems are always up to date and only occasionally have problems
If you think running without security updates is a good thing than that is your problem.
### For people who want new0ish software but don't like the idea of running something "unstable" there's the testing branch.
People want *some* new software (say a new Gimp or KDE or whatever they need most for they daily work), they don't want to have a completly bleeding edge that doesn't even provide security updates.
### How are we supposed to have the time/energy/desire to build binaries for all these arches, much less hardware to test them on?
For the beginning standard source format would already help a lot, ie. one with computer readable metainformation on how to build, not just human readable INSTALL files. The rest could then be simply autobuilded.
### but there are good technical reasons why it can't be as simple as we'd like.
I wouldn't call it 'good technical reasons', most of it is simply historic and lack of standardization. Nothing would really stop a standard package format to work out on a large scale if people simply tried a little bit harder.
### You definately should not be able to install it system-wide no matter what package management system you have.
Of course you should not, unless you are root, thats now something bloody obvious that I didn't assume that it would be needed to mentioned specially. What I was talking about is installing stuff in the users home directory.
### If you want a cutting edge version of gimp, download and compile it in your home directory.
Why should the user be forced to compile and track dependency manually when the Debian repository provides a perfectly working version of it? Why can't apt-get allow him to install it into his home? Weren't package managers invented to avoid compiling from source in the first place?
### Sure debian stable is a little behind but remember, it's *stable* meaning that buggy cutting edge program won't be there.
You don't seem to have used Debian for too long, 'stable' refers to the package dependencies *not* to the software. There is plenty of software in stable that is far more 'bleeding edge' than unstable, simply because 'stable' was frozen at a time where the software just has jumped to a new major version and never got any of the bug fixes that the software in unstable got. After all only security fixes get down to stable, rest stays in testing and unstable. Anyway, thats a different issue.
### but I don't expect the author of a program to package their program up for each distro.
Neither do I, what I want is a way to package programms independly of the distro. Which either means source tar.gz + some computer readable metadata (as oposed to human readable INSTALL files like its handled at the moment) or autopackage/lsb-rpm style binaries.
### Why do you think that it's a sensible idea for users to be installing software in the first place?
Do you think its a better idea to run as root for every little piece of software you want to try out? What makes you think that its actually not a sensible idea to install software as user? Its a perfectly fine thing todo and its the way Windows and MacOS work for decades. You won't stop viruses from spreading by forcing users to either 'sudo' or use untrusted sources instead of official Debian repositories.
### In which case you can simply build the software to work in that environment.
But weren't package managers invented to avoid that in the first place?
### That's not what apt-get is supposed to do.
Now thats a great argument. If that is really the case I would call that broken by design, but could you care to give some details why exactly it wasn't supposed to handle such a situation? After the only difference between installing as root and installing as user is the $PREFIX, something reasonably easy to fix if one tries to. There are of course packages that have to be installed as root (kernel and such), but they are by far the minority.
### You can install software in your home directory in Linux.
Yes, you can. However the distros packagemanager and the distros package repository is completly useless for doing so and thats exactly the point of why apt-get isn't a very good solution for handling software. It works great for some uses, but completely fails at others.