Dear lord, no. Bad & outdated third-party documentation is the software world's biggest usability problem.
Software packages should *include* the upstream documentation. That way, the user gets correct documentation that matches the version of the software they installed. If the documentation is very large, it can go into a separate foo-doc package.
The other advantage is that people using the software offline can access its documentation.:)
Unfortunately, the only real solution to this is source based installs or having a compile for every combination of dependencies.
There is another alternative. Debian prefers to split the package up.
For instance, you would have a foo package, containing/usr/bin/foo, and its LDAP-specific functionality would be in the foo-ldap package that contains/usr/lib/foo/plugins/libfoo-ldap.so.
There is no need to compile the package multiple times with a different build configuration, unless the package doesn't use a shared library/plugin system. These days such packages are rare, and if the code base is of half-decent quality then modifying it and sending the patches upstream is not difficult.
Is it really hard to run dpkg-deb --build? You can even build a deb with ar if you really want, you would just have to tweak the magic number at the top of the file once you're done.
It is no less standardised, but more likely to be blocked, than a random port. If you are going to avoid TLS then you may as well just use a randomly chosen port.
If you care about privacy then you will be using PGP, S/MIME or another end-to-end encryption system.
If you really don't trust your ISP to not read your mail then using your own server is not enough--they (or anyone else whoses systems your mail goes through) can use something like tcpdump to capture all your mail and other traffic.
While the original poster probably had no idea that POSIX ACLs existed, he is still correct when he says that the ACLs in Windows NT are more powerful than POSIX ACLs.
You're talking about Mandatory Access Control, such as implemented by SeLinux? You didn't make that terribly clear in your post. Besides, I didn't know Windows had any form of MAC. Its entire security model is based on Discresionary Access Control, like most Linux-based operating systems.
It is just as much Microsoft's fault for sticking with the assinine default configuration (users are admins by default) that made it unnecessary for Apple to ever learn how to code properly.
Great post, very informative summary of what can no doubt be extracted from many hours spend in the technote library. If only GNU/Linux distributions could learn from Apple's example, instead of simply wrapping sudo with a GUI popup.:(
Everything is moving to i386 and amd64. Now that Apple no longer produces ppc machines, is there an easy way for a regular user to get their hands on one?
Users who care about freedom could also just check out Debian in the fist place. I don't really undestand the point of gNewSense when Debian already exists, it seems like wasted effort.
I know of all of that. MICROS~1's failure to get ISVs to write decent applications is a direct consequence of their failure to implement proper security (that is, all users are not admins) out of the box with Windows NT/2000/XP (choose one depending on when you think this issue started to be important. My choice is MS-DOS 1.0).
It sounds like you have already practically agreed with me that MS putting their foot down and making regular users not be administrators by default would have been the only way to make ISVs code their programs properly.
UAC and the virtual per-user folders are just more hacks and band-aids applied to a fundamentally flawed design.
Slashdot Mirror
dpkg --root=$HOME --install foo.deb
That option also makes it easy to set up chroots.
Dear lord, no. Bad & outdated third-party documentation is the software world's biggest usability problem.
:)
Software packages should *include* the upstream documentation. That way, the user gets correct documentation that matches the version of the software they installed. If the documentation is very large, it can go into a separate foo-doc package.
The other advantage is that people using the software offline can access its documentation.
Is it really hard to run dpkg-deb --build? You can even build a deb with ar if you really want, you would just have to tweak the magic number at the top of the file once you're done.
Wow, how did the Internet get invented, since it was around before the development of the 233MHz Pentium processor!?
Take them to small claims, for the cost of buying your domain back from the scumbag squatter + your time.
Sounds like that costs more than taking them to small claims.
What is wrong with "approved", "accepted" or (since the editors are those who actually posted the stories) "posted"?
It is no less standardised, but more likely to be blocked, than a random port. If you are going to avoid TLS then you may as well just use a randomly chosen port.
If you care about privacy then you will be using PGP, S/MIME or another end-to-end encryption system.
If you really don't trust your ISP to not read your mail then using your own server is not enough--they (or anyone else whoses systems your mail goes through) can use something like tcpdump to capture all your mail and other traffic.
That runs the entire process as root however--hardly the same thing as UAC.
While the original poster probably had no idea that POSIX ACLs existed, he is still correct when he says that the ACLs in Windows NT are more powerful than POSIX ACLs.
You're talking about Mandatory Access Control, such as implemented by SeLinux? You didn't make that terribly clear in your post. Besides, I didn't know Windows had any form of MAC. Its entire security model is based on Discresionary Access Control, like most Linux-based operating systems.
None of these are something I would run Ubuntu on. I would run Debian on them. :)
It is just as much Microsoft's fault for sticking with the assinine default configuration (users are admins by default) that made it unnecessary for Apple to ever learn how to code properly.
Great post, very informative summary of what can no doubt be extracted from many hours spend in the technote library. If only GNU/Linux distributions could learn from Apple's example, instead of simply wrapping sudo with a GUI popup. :(
Everything is moving to i386 and amd64. Now that Apple no longer produces ppc machines, is there an easy way for a regular user to get their hands on one?
Feel free to use my letter as a starting point for your own: http://robots.org.uk/nvidia.shtml
Good answer. It is also compatible with Ubuntu's current behaviour, which they reaffirmed with this news article. :)
Binary blobs running in the kernel that are impossible to debug, alter or improve would not make a better release.
Could you link me to somewhere that will sell me one of these cards?
Because it's *so* hard to install the nvidia-whatever packages from the package archive.
You should forget about reading third party howtos from mailing lists and wikis. Just use the distribution's own documentation and packages.
Users who care about freedom could also just check out Debian in the fist place. I don't really undestand the point of gNewSense when Debian already exists, it seems like wasted effort.
I know of all of that. MICROS~1's failure to get ISVs to write decent applications is a direct consequence of their failure to implement proper security (that is, all users are not admins) out of the box with Windows NT/2000/XP (choose one depending on when you think this issue started to be important. My choice is MS-DOS 1.0).
It sounds like you have already practically agreed with me that MS putting their foot down and making regular users not be administrators by default would have been the only way to make ISVs code their programs properly.
UAC and the virtual per-user folders are just more hacks and band-aids applied to a fundamentally flawed design.
What has that got to do with the packages of the NVIDIA driver?