Yeah? I could redesign the boot system on my system in an hour to do that -- it's really simple. But you know what? I'm not. I like knowing when something's failing.
Hmm...yeah, I'll give you that. I don't think that what you're claiming is all that true for gui apps, but in the console world, there are some egregious issues.
perl? grep? touch? May be accurate acronyms, but don't help users much.
less? Probably seemed like a cute idea at the time.
umount? This was *not* worth saving a single character for.
Don't use sendmail unless you really, really know that you definitely need to use sendmail. If you're running a single workstation or anything below a very large ISP, postfix is a really nice alternative. It actually has sane defaults and a sane configuration interface, which makes it easy to set up and means that you're unlikely to have overlooked holes somewhere.
Actually, *most* programs maintain only a man page. A few (FSF stuff mostly) maintain only an info page. A few maintain both.
I really don't like the info interface -- i'd rather be using lynx or links if I need to be traversing hypertext.
I'd rather obsolete info, have a single markup standard that can translate to man or html (if there isn't already something in groff or tex or xml or something for this), and then let the installing system generate documentation of the desired type at install time as part of the build process.
And MS isn't worried about going out of business. If there's a viable alternative to their product, they have *much* less leverage to squeeze companies. So even a few percent of marketshare hurts, and Linux has a good chunk of the server market.
OTOH, while I'm sure there are folks at MS responsible for dealing with desktop threats from Linux, countering desktop Linux can't possibly be a priority at the minute.
I agree that C-A-Esc in KDE is as reasonable as C-A-Del is in Windows.
However, emacs in text fill mode is not soft wrapping. It does indeed wrap things on screen, but it also leaves the linefeeds in the resultant file. That's hard wrapping.
Ever tried ctrl-alt-escape in KDE Obvious and transparent, no?
I'd say slightly more obvious than Window's ctrl-alt-delete, which I don't see hordes of users complaining about.
So because some people stupid, things should be made much more complicated than they need to be for everyone else?
UNIX rm doesn't ask for confirmation. Deletion in MS Explorer does. Formatting and system-modifying commands are tucked away in corners. Should it be trivial to open a big security hole in your system? I'd say no.
[long bit about how to change resolutions in Windows]
And this requires significantly more steps and time than in Linux. It's like like holding down shift to select a range is in Windows. Yes, it's a key that someone has to tell you about, but it isn't exactly hard to use or remember.
So the problem is that you have to be root to access/dev/lp0? Just change the permissions on/dev/lp0. Make a group "printers", add yourself to it, and chgrp printers/dev/lp0, and then chmod g+rw/dev/lp0.
I guess/sbin/reiserfsck is just a softlink to/bin/true, then?
I don't use ReiserFS -- after my roommate had some unpleasant experiences with it, I've become a happy ext3 user. However, my guess is that it does nothing more than replay the journal -- evidently there's a userspace app that does this work instead of the driver doing it. If so, this cannot fail -- journalling filesystems are designed so that a power loss at any point will keep a valid journal (though the journal might not be as complete as you like...the last file you saved may not actually have been written to disk yet).
What's wrong with a simple, Windowish config dialog that creates a printcap for you?
Nothing. As I said, if this is just a frontend, I'll give him a pass. My greatest fear is that someone is going to tie the actual "next-gen" printing system to GNOME and/or KDE. Maybe you have binary config files, I dunno. The print system is optional, so old-timers can still use lpr, but lpr is slowly used less and less until you start relying on the new system, and hence KDE/GNOME. Look at programs like gtk-gnutella, lopster, and Downloader for X. All of these perform tasks that in a traditional UNIX environment should be performed by a daemon. It should be possible to interact with each via a command line interface, and if a GUI frontend sits on top...well, that's just peachy. However, instead all three have taken the "Windows route" of assuming that a GUI will always be present and forcing one to rely on it. I'd hate to see printing do the same.
I think this is just a misunderstanding. You think of daemons as apps, the original author thinks of daemons as part of the system
This is a good thought, but the programs that I was talking about would likely run into problems. xbindkeys displays no window, but opens an X connection to a server. If you require that an "app" be something that displays a window, I worry about processes that normally display nothing but could potentially pop up an error dialog. I grant that this is more common under Windows than Linux, but they certainly exist. Now, you *could* layer this on top...say, make all GNOME or KDE apps register themselves as "regular apps"...but besides adding complexity, you'd never get every author to follow this, which would produce inconsistent behavior from a user point of view.
As for killing bad Netscape Navigator instances...well, NN is a closed source program that there was once no good alternative for, so you couldn't simply "fix the program" as I suggested earlier. This is a bit of a special case, particularly since I know of no other app besides NN that goes haywire like this. I think I'd rather have a cron job that kills NNs that have been runnable constantly for the last N minutes, or some othe application-specific solution.
Absolutely right...if you actually know this, and if you have configured X accordingly, i.e. more than one mode.
Okay, you do have a good point here. It's true that Linux distros don't do a good job of telling new users about C-A-KP+ and C-A-KP-. Maybe a help system (if GNOME and KDE help systems ever get up to par) will list that.
OTOH, even if XFree86 4.x doesn't properly detect your settings (which I've had no problems with), at least Xconfigurator generates multiple modes. I haven't used any other front end, so I can't vouch for anything else.
I will grant you that I'm pretty darn likely to hit okay for all the confirmation messages that I've ever seen from fsck, but then I've also recovered vital emails from a unrepairable filesystem.
When it comes to data integrity, there isn't much screwing around you want to do. If there's a chance that it's all going to die, warn someone.
Add -a to the fsckoptions. If it still complains, try using yes|fsck -a in your initscripts.
If that still doesn't suit you for some reason, then get upset.
If I had a single CD that contained all the source I'd written before, you'd be darn sure that I'd dump an image of the damaged filesystem before trying to repair it. This gives me that option.
Forcing you to change your desktop size when you want to change resolution to play a game or watch a movie is something that Windows forces you to do. It's poor Windows design, but people have gotten used to it.
XFree86 does this right. You set the desktop resolution to the highest resolution that your monitor/video card can do. If that's "too small" then you increase font sizes. Decreasing the resolution and wasting what your hardware can do is *not* the answer.
As for being familiar with tech, there's a profound difference between the MS and Apple help systems.
Apple's help systems are exemplified by Apple Help. You want to do foo? The system will show you how to do bar.1, bar.2, bar.3, and the other steps to accomplish foo. After finishing, you understand what happened and how in the future to accomplish foo, perhaps with custom changes.
OTOH, MS's help systems are based around wizards. You want to do foo? You can have a wizard do foo for you, but you have no idea about how to do bar.1, bar.2, bar.3, etc. You just know how to choose from a list of preselected tasks. If you had to perform a variant of one of these tasks after years of using the wizard, you'd still have no idea what to do.
The Apple/MS dichotomy is something like the "give a man a fish" truism. If you perform a computing task for a user, you help him for the minute. If you *teach* a user how to perform a computing task, you help him for the rest of his life.
Freenet may not be perfectly anonymous, but it's very, very difficult to break. OTOH, Gnutella is incredibly easy to track people on and build up incriminating logs...and yet, the MPAA took years to get around to trying to work it over.
Slashdot Mirror
aspell is pretty good. I use it from within emacs quite happily. Also, I can just tap meta-$ to check the previous word if I'm unsure of a spelling.
Yeah? I could redesign the boot system on my system in an hour to do that -- it's really simple. But you know what? I'm not. I like knowing when something's failing.
I mean how many people are not going to want to use fonts that look nicer
blurrier
Hmm...yeah, I'll give you that. I don't think that what you're claiming is all that true for gui apps, but in the console world, there are some egregious issues.
perl? grep? touch? May be accurate acronyms, but don't help users much.
less? Probably seemed like a cute idea at the time.
umount? This was *not* worth saving a single character for.
Don't use sendmail unless you really, really know that you definitely need to use sendmail. If you're running a single workstation or anything below a very large ISP, postfix is a really nice alternative. It actually has sane defaults and a sane configuration interface, which makes it easy to set up and means that you're unlikely to have overlooked holes somewhere.
Actually, *most* programs maintain only a man page. A few (FSF stuff mostly) maintain only an info page. A few maintain both.
I really don't like the info interface -- i'd rather be using lynx or links if I need to be traversing hypertext.
I'd rather obsolete info, have a single markup standard that can translate to man or html (if there isn't already something in groff or tex or xml or something for this), and then let the installing system generate documentation of the desired type at install time as part of the build process.
Really? I know one person that works at Microsoft, and he runs Linux as well. As a matter of fact, he keeps a Tux in his cubicle.
Are there actually any coders that *like* Windows at MS?
Linux. Is. Inconsequential.
Not in the server market, it isn't.
And MS isn't worried about going out of business. If there's a viable alternative to their product, they have *much* less leverage to squeeze companies. So even a few percent of marketshare hurts, and Linux has a good chunk of the server market.
OTOH, while I'm sure there are folks at MS responsible for dealing with desktop threats from Linux, countering desktop Linux can't possibly be a priority at the minute.
I might disagree with what the top 10 problems are (a lack of freecell wouldn't be very high on my list)
PySol beats the snot out of freecell.
I agree that C-A-Esc in KDE is as reasonable as C-A-Del is in Windows.
However, emacs in text fill mode is not soft wrapping. It does indeed wrap things on screen, but it also leaves the linefeeds in the resultant file. That's hard wrapping.
Ever tried ctrl-alt-escape in KDE
Obvious and transparent, no?
I'd say slightly more obvious than Window's ctrl-alt-delete, which I don't see hordes of users complaining about.
So because some people stupid, things should be made much more complicated than they need to be for everyone else?
UNIX rm doesn't ask for confirmation. Deletion in MS Explorer does. Formatting and system-modifying commands are tucked away in corners. Should it be trivial to open a big security hole in your system? I'd say no.
[long bit about how to change resolutions in Windows]
And this requires significantly more steps and time than in Linux. It's like like holding down shift to select a range is in Windows. Yes, it's a key that someone has to tell you about, but it isn't exactly hard to use or remember.
And C-A-Del is obvious in Windows?
At least Alt-F4 is listed on menus, so a user might see the listing, but it sure doesn't seem all that intuitive to me.
So the problem is that you have to be root to access /dev/lp0? Just change the permissions on /dev/lp0. Make a group "printers", add yourself to it, and chgrp printers /dev/lp0, and then chmod g+rw /dev/lp0.
I guess /sbin/reiserfsck is just a softlink to /bin/true, then?
I don't use ReiserFS -- after my roommate had some unpleasant experiences with it, I've become a happy ext3 user. However, my guess is that it does nothing more than replay the journal -- evidently there's a userspace app that does this work instead of the driver doing it. If so, this cannot fail -- journalling filesystems are designed so that a power loss at any point will keep a valid journal (though the journal might not be as complete as you like...the last file you saved may not actually have been written to disk yet).
What's wrong with a simple, Windowish config dialog that creates a printcap for you?
Nothing. As I said, if this is just a frontend, I'll give him a pass. My greatest fear is that someone is going to tie the actual "next-gen" printing system to GNOME and/or KDE. Maybe you have binary config files, I dunno. The print system is optional, so old-timers can still use lpr, but lpr is slowly used less and less until you start relying on the new system, and hence KDE/GNOME. Look at programs like gtk-gnutella, lopster, and Downloader for X. All of these perform tasks that in a traditional UNIX environment should be performed by a daemon. It should be possible to interact with each via a command line interface, and if a GUI frontend sits on top...well, that's just peachy. However, instead all three have taken the "Windows route" of assuming that a GUI will always be present and forcing one to rely on it. I'd hate to see printing do the same.
I think this is just a misunderstanding. You think of daemons as apps, the original author thinks of daemons as part of the system
This is a good thought, but the programs that I was talking about would likely run into problems. xbindkeys displays no window, but opens an X connection to a server. If you require that an "app" be something that displays a window, I worry about processes that normally display nothing but could potentially pop up an error dialog. I grant that this is more common under Windows than Linux, but they certainly exist. Now, you *could* layer this on top...say, make all GNOME or KDE apps register themselves as "regular apps"...but besides adding complexity, you'd never get every author to follow this, which would produce inconsistent behavior from a user point of view.
As for killing bad Netscape Navigator instances...well, NN is a closed source program that there was once no good alternative for, so you couldn't simply "fix the program" as I suggested earlier. This is a bit of a special case, particularly since I know of no other app besides NN that goes haywire like this. I think I'd rather have a cron job that kills NNs that have been runnable constantly for the last N minutes, or some othe application-specific solution.
Absolutely right...if you actually know this, and if you have configured X accordingly, i.e. more than one mode.
Okay, you do have a good point here. It's true that Linux distros don't do a good job of telling new users about C-A-KP+ and C-A-KP-. Maybe a help system (if GNOME and KDE help systems ever get up to par) will list that.
OTOH, even if XFree86 4.x doesn't properly detect your settings (which I've had no problems with), at least Xconfigurator generates multiple modes. I haven't used any other front end, so I can't vouch for anything else.
I will grant you that I'm pretty darn likely to hit okay for all the confirmation messages that I've ever seen from fsck, but then I've also recovered vital emails from a unrepairable filesystem.
When it comes to data integrity, there isn't much screwing around you want to do. If there's a chance that it's all going to die, warn someone.
Add -a to the fsckoptions. If it still complains, try using yes|fsck -a in your initscripts.
If that still doesn't suit you for some reason, then get upset.
If I had a single CD that contained all the source I'd written before, you'd be darn sure that I'd dump an image of the damaged filesystem before trying to repair it. This gives me that option.
Actually, I had fun with an old S3 video card periodically hard-freezing my system in Windows NT.
Or those who use VT100s to do all their work?
Use main memory if you run out of video memory.
This isn't a bug -- it's a design feature.
Forcing you to change your desktop size when you want to change resolution to play a game or watch a movie is something that Windows forces you to do. It's poor Windows design, but people have gotten used to it.
XFree86 does this right. You set the desktop resolution to the highest resolution that your monitor/video card can do. If that's "too small" then you increase font sizes. Decreasing the resolution and wasting what your hardware can do is *not* the answer.
Inquiring minds want to know.
As for being familiar with tech, there's a profound difference between the MS and Apple help systems.
Apple's help systems are exemplified by Apple Help. You want to do foo? The system will show you how to do bar.1, bar.2, bar.3, and the other steps to accomplish foo. After finishing, you understand what happened and how in the future to accomplish foo, perhaps with custom changes.
OTOH, MS's help systems are based around wizards. You want to do foo? You can have a wizard do foo for you, but you have no idea about how to do bar.1, bar.2, bar.3, etc. You just know how to choose from a list of preselected tasks. If you had to perform a variant of one of these tasks after years of using the wizard, you'd still have no idea what to do.
The Apple/MS dichotomy is something like the "give a man a fish" truism. If you perform a computing task for a user, you help him for the minute. If you *teach* a user how to perform a computing task, you help him for the rest of his life.
Or use a front end.
Freenet may not be perfectly anonymous, but it's very, very difficult to break. OTOH, Gnutella is incredibly easy to track people on and build up incriminating logs...and yet, the MPAA took years to get around to trying to work it over.
It's the lack of source anonymity that makes me hold off on hosting the files I've acquired.
Want anonymity? Try Freenet.
Beautiful post.