Sun did make the mistake of assuming everyone had their own domain, but I think they were assuming only companies and similarly sized organizations would be writing anything much bigger than an applet. You could always have used nz.email.lpg_name for your previous project - it doesn't really matter, as long as nobody else would want that package name.
This scheme is intended for library-type classes; I suspect the "executable" for many Java projects is in the "null package" (i.e. the project's root directory).
When I said "most significant" I meant it in the sense of "most significant bit in a byte". If I write two related classes called SuperWidget and UberWidget which do stuff, and you write classes with the same names which do different stuff, it makes sense for my widgets to be in the same place in the directory tree, and your widgets to be somewhere else in the directory tree, rather than putting our (possibly unrelated) SuperWidgets in the same folder because they happen to share a name.
I agree, for small-scale or amateur developers, this scheme isn't ideal - but it really doesn't matter what the exact package name is, as long as nobody else will want it. It's like XML namespaces - an XML namespace is represented by a URI (usually an http:// one), but there's no need for there to actually be anything at that address (some people put the DTD or XML schema there, but the W3C actually discourage that because they don't want people to get used to the idea that namespaces are anything except an abstract identifier). They're only using domains and URIs at all in order to borrow an existing unique identifier, instead of having to implement some sort of separate registering system.
The global namespace thing is just that each package's full name includes an identifier for the server it's on.
Here's an analogy:
http://slashdot.org/comments.pl?sid=27540&cid=29 58 559
There's a uniquely identified item in a global namespace. I might use a mirror (web proxy) or an untrustworthy source (r00ted web proxy) to access it; that's my problem.
/comments.pl?sid=27540&cid=2958559
There's an item in a local namespace. Exactly the same as the one above, but
you need to know the context (i.e. that we're on Slashdot) before you can actually retrieve it.
If you decide to download, for example, a snapshot of gcc@gnu.org--gcc--3.0 (as far as I can see, that's roughly what gcc 3's unique ID might look like if the authors used arch) from l33t.h4x0r.net instead of ftp.gnu.org, that's your security problem, not the identifier's. If you don't think the site you're connecting to is the canonical source, don't use it - just like you'd download gcc-3.0.tar.gz (or whatever the distribution filename is) from a reputable mirror rather than l33t.h4x0r.net at the moment.
The canonical package name for your widgets would be nz.net.neural.(anything you like here)
If you own multiple domains (subdomains, or not), you pick one or more to use. The most sensible strategy would be to pick the one you were most likely to keep. Whether it corresponds to a real web page, or server, or whatever really doesn't matter - all that matters is that you control the neural.net.nz domain, and you don't use the same package name for different things as anyone else at that domain.
You do use directories for package name components - the class file for nz.net.neural.widgets.Widget (the convention is for class names to have initial caps) should go in nz/net/neural/widgets/Widget.class (replace / with your OS's directory separator if you don't use Unix). You often don't see this because classes are in.jar files, which have their own internal directory structure (they're slightly modified zip files).
The domain has to be written backwards to put the most significant part first (otherwise neural.net and neural.net.nz would have overlapping namespaces, even though they might be owned by different people).
When did your Windows box last upgrade its kernel, then?
So Windows Update consists of DLL upgrades (or shared library upgrades, to use the Linux terminology) - a lot like, say, an APT update on Debian, or whatever the equivalent's called on Red Hat.
If you want to be messing with custom-compiled kernels, that's your choice, and it's not as trivial as updating system libraries and programs. But then it's not a choice Windows even gives you, so I can't say I'm complaining:-) From what I've gathered about Red Hat, Windows has much the same approach as a stock install of RH, in fact - you get a standard, intensively-tested kernel which isn't exactly on the cutting edge, but is stable.
One thing Microsoft's strategy does have in its favour is that kernel-specific device drivers aren't a problem when all the copies of, say, Win2k have the same kernel. However, don't some Linux distros distribute packages of kernel modules compiled for their standard kernel (like the modified 2.4.9 kernel Red Hat use)?
That's exactly what/usr/local is for - locally compiled software. On most GNU and GNUish software the author sets it up to install to/usr/local by default, but you can do./configure --prefix=/usr if you're building a distro package.
I don't know what other distros are like about this (I've only ever used Mandrake and Debian, and I didn't get experienced enough with Mandrake to know any of the internals), but Debian source packages come in two parts - a tarfile of original, unmodified source, plus a.diff.gz file containing the changes ("Debianizations") the Debian package maintaniner made to make it fit in with Debian conventions (moving all documentation to/usr/share/doc/name-of-the-package, for instance). If the original author's makefile or other code doesn't conform to Debian conventions, the maintainer will change it so it does.
For a program like you describe where (presumably)/usr/local/bin is hard-coded somewhere, the diff would include replacing that with/usr/bin - you, as an "upstream" developer, can probably make this easier by defining PREFIX to/usr/local and always referring to "$PREFIX/bin" and so on.
The things behind dpkg and all of that sure, they are nice, but how are they all that really different from RPM? I agree that I may not be totally clued in about all aspects of apt, dpkg and all of the other stuff Debian uses, but to say that it's so tied to a packaging format??? Debian is no more tied to it then Redhat is tied to RPM. If they are, then the Debian project made a major mistake!
Debian's primary advantage is that its packages are consistent. When you select a package, it comes with a very specific set of dependencies. That, and the fact that Debian is the source of the vast majority of.deb packages (especially the sort of "core" packages likely to come up in a dependency), mean they can get away with a lot of splitting up components, compiling platform-independent bits once and binaries many times (Debian isn't just an i386 distro like many others), shared libraries that get shared properly, and all sorts of other consistency stuff. Where a release-focused distro might concentrate on making all the packages in version 9.2 (say) work together nicely and consistently, Debian does a pretty good job of keeping all the packages consistent, all the time. (OK, so it doesn't always work 100% on 'unstable', but if you wanted stability you wouldn't be running that version anyway).
Maybe now is the time for Debian to actually form a company or form a different way of making the decisions instead of democracy.
The democracy/meritocracy structure is exactly what Debian is about - it's run by a non-profit org. called Software in the Public Interest. They don't exist to make a profit, or to make themselves popular - they exist to make a distribution of good software freely available.
Maybe they need to modify the DFSG to be more lienient?
And that would help how?
If it's free (free as in DFSG), it goes in 'main' and goes on the official CDs. If it's free as in DFSG but requires non-free software (like a GPLed "helper" application for PGP for instance), it goes in 'contrib'. If it's not free-as-in-DFSG (this includes shareware and closed-source freeware), it goes in 'non-free' - contrib and non-free are easily available over ftp or as the last CD in the set. If it's not in non-free, either nobody's tried packaging it, or Debian aren't sure if it would even be legal for them to distribute it.
Again, things like the DFSG are why Debian exists.
I don't know, but there has to be someone who is going to draw a line in the sand and get the volunteers in action and get Woody released (and with a 2.4 kernel as well...).
It looks like it'll be installed with a 2.2 kernel by default, and a 2.4 kernel as an option (fairly easy to install thanks to the magic of apt/dpkg - install, reboot, Lilo offers you a choice of kernel). Woody already includes precompiled 2.4.17 kernels for 386, Pentium, 686 (PPro/P2/P3/Celeron), K6 and K7 (Athlon/Duron), and that's just the i386 builds.
OTOH, if Debian was a company, where would all those volunteers be? Probably starting their own distro or doing Linux from Scratch.
PHB's and Joe Sixpacks like hearing and seeing commercials like Mandrake Linux is compliant (it isn't but) with the LSB which means no matter where you buy or download your software it will work!
This is why volunteer efforts like Debian have a niche. They don't like hearing and seeing commercials, they like giving people a stable OS. They're not trying to make a profit, which is why they can get away with doing what they feel is The Right Thing.
Re:MS core fonts are free for noncommercial use
on
Scalable-Font Tools?
·
· Score: 2, Informative
while a distro that provided an installer script would be fine.
Indeed, Debian does precisely that as part of its 'contrib' archive. (As do other distros, probably)
OK, so it's his mailing list, and his e-mail, so he has a right to do this. Even so:
Should self-proclaimed free software advocates really lower themselves to Microsoft's level? I'm sure this sort of elitism hurts the free software cause more than it helps it.
"I can't read your badly-formatted documents, so I'll deliberately stop your badly-written mail client reading mine.:P"
Sound childish to anyone?
Isn't the way to promote free software to be better than the competition? Microsoft can get away with assuming everyone uses Windows + Outlook Express, because we already know they're arrogant. If you claim to represent free software, being that arrogant risks having free software labelled as arrogant and elitist too.
Yes, RMS was right to be critical of using Word docs as attachments (I don't always agree with RMS's opinions, my opinion tends to be somewhere between open source and FSF). Critical is fine. If asked, most people on/. would be critical of Microsoft Out-of-order. But I can't really agree with deliberately breaking it.
Yes, IE 4 supports at least the most commonly used bits of CSS. NS4 is much worse (bizarre paragraph spacings and overlapping text and images were the main problems my site had on it).
Luckily there's a related Netscape bug. If you import your stylesheets with
@import url(/styles/mystyle.css);
instead of using the tag, Netscape 4 will completely ignore them, meaning NS4 users get a plain black-on-grey unstyled site - not too pretty, but at least they can read it without bits of it overlapping.
Windows limits users' capabilities by user interface as well as by permissions. If you have sysadmin-type priviledges (which you always do on Win9x) you can probably do raw hardware access (I'm thinking hard disks here), but not easily. You can format disk partitions, but you'll always be asked to confirm it. You can only make and delete partitions by rebooting to DOS mode (at least on Win9x, I'm not sure about NT-based versions like 2k and XP).
In Linux (as far as I can see) the user is assumed to be right, as long as they have sufficient priviledges. (Think "cat/dev/zero >/dev/hda", which will cheerfully fill your partition tables and disk with zero bytes). This is great if you know what you're doing, which is why Linux users are encouraged to always use a mere mortal userID, and only switch to root when they know exactly what they're doing (well, what they're about to do, anyway).
This is why running as an admin-level user on Windows is pretty common, while running as root on Linux is, er, frowned upon.
For example: I know someone who works from home and occasionally goes in to the office, and uses the same (company-owned) laptop in both places. Because he needs to be able to alter network settings to switch between work LAN and home cable, the company's sysadmin (i.e. the guy with the laptop's Administrator password) has added admin-level access priviledges for that laptop to the login he usually works in.
If that was a Linux box, I assume the sysadmin would add the user to the list of sudo users, or even just give him the root password; either way, he'd have to make a conscious decision "I need super-user access for this; I'd better be careful".
Pseudorandom number generators such as the one used by rand() are predictable. AFAIK,/dev/random is extremely unpredictable because it uses hard-to-reproduce external data as an extra source of randomness - IIRC, it uses the precise (nearest millisecond?) timing of interrupts from the network cards, keyboard, mouse, etc. Not something you'll be able to replicate easily!
1) Notice it says "Can the developer of a program who distributed it under the GPL later license it to another party for exclusive use?" - there's nothing wrong with licensing it to another party for non-exclusive use. Look at the license on Cygnus Solutions' Cygwin DLL, which provides glue code to emulate a sensible Unix environment on Windows. Cygnus (now owned by Red Hat) dual-licenses it - the first is GPL (with extra permission to link to any OSI-approved open source software) and is free (as in free beer), the second is closed (but you can legally link it to a non-free app) and costs money. Borland are doing this too, with Kylix - the free (beer) downloadable version comes with GPLed runtimes, so you can only legally compile GPLed software on it, but the commercial version comes with the same libraries under Borland's usual "once you've paid us, you can link it to anything" licensing.
2) I suspect this is dodgy ground legally, and depends where X and Y live. If Y sends a few lines of diffs, they're probably not significant enough to be covered by copyright (your laws may vary). If Y explicitly puts their contributions in the public domain, X can claim it as their own and do what they like with it (that's what public domain means). If Y explicitly signs over copyright on their contributions (like the FSF requires on contributions to GNU code), X now owns it and can do what they like. Otherwise, X could probably argue that Y had implicitly given them permission to use their contributions... I think your quote is probably right in this case, but it seems like dodgy legal ground, and IANAL.
Focus-stealing when a new window appears depends on your window manager. The one built into KDE (does it have a name, or is it just "KDE"?) probably always steals focus. Sawfish can if you want it to (it does by default, I think; Customize->Focus to change its behaviour). Those are the only ones I've used a lot, so YMMV with others.
Are we benefiting MS by using Hotmail, though? My guess is the millions of disposable spam accounts they host soak up bandwidth and/or disk space without Microsoft actually gaining anything from the whole thing (at least on balance).
Yeah, OK, I meant it's not called the stable distribution:-) and there aren't official ISOs. I hadn't realised there were Sid ISOs too, though...
(I'm using unstable at the moment, the only serious instability I've noticed is the nvidia X driver which Debian have nothing to do with)
2.2r4 is the latest official CD-image. It's potato release 4, not woody (potato = 2.2, woody = 3.0).
When Debian say "the stable version", that's what they mean... their aim is to get each release to all work together nicely, and they don't release that often.
woody contains XFree86 4, gcc 3 and other newer goodies, but it hasn't become the stable distribution (it's still classified as testing) so there are no official ISOs.
They don't (or try not to) do ISOs as such, but the recommended way to make a Debian ISO is to download the packages to compile a "pseudo-image", then use rsync to turn that into the real ISO. They've done a convenient "pseudo-image kit" for Unixes and Windows (AFAIK, it may do more platforms) which consists of executables (probably source for the Unix version, I don't know since I used the Windows one), a readme, and pointers to where to get package mirrors, rsync mirrors and package lists from.
Using the pseudo-image kit (even on Windows) is at least as easy as actually installing Linux (read instructions, follow instructions - I get the impression Debian is aimed at people who can read instructions, so that suits them fine).
The resulting CD is bootable and works fine (and once you've finished with it, its bootability, built-in root FS and utilities make it a nice rescue disk).
Basically, with the GPL under copyright, you can say "this software is licensed under the GNU GPL" and it's unambiguous what the terms are.
This isn't necessary for, say, BSD-style licenses (BSD, ZLib, LibPNG) because they're simpler and shorter - it's reasonable to include the entire BSD license in each file of your source, so you don't say "this is BSD-licensed", you say "you may do this, this and this but not this". "BSD license" is just a convenient shorthand for describing things - but from a legal point of view, the license consists of a couple of paragraphs embedded in each source file.
However, it's obviously not reasonable to include the whole GPL in the same way. The GPL is long (20K?) because of copyleft - it's less permissive than the BSD license, so it can't just grant blanket permissions like the BSD license does (although an abbreviated GPL without the preamble/manifesto would be nice, since they're not really part of the license as such).
If the GPL was free (in the FSF sense of the word) or open source, you'd get people redefining what it meant, and much confusion would ensue. ("Our software is licensed under the GPL." "No it isn't, ours is the real GPL!")
Network cables and some business phone jacks, however, are physically compatible and can be indistiguishable. I have made that mistake, and you would as well. Blame industry for it: choosing a phone cable standard for Ethernet was NOT a smart move.
I can agree with that. Each student room in the college I'm at has a double socket on the wall - one RJ45 network connection, and one physically identical phone socket (which you need an adaptor to plug a normal phone into). All we were told was that the network connection was *usually* the one on the left:-)
It's interesting that in The Feeling of Power, the guy who rediscovers arithmetic does so by reverse-engineering old obsolete computers and slowly working out how they do it:-)
They're not perfect. I don't get crashes that often, but occasionally (once every couple of days, usually while in a game but sometimes just randomly) X freezes and won't even respond to keyboard input. Luckily I have a palmtop (an old Psion 3c) and a serial cable, so I can still log in on a serial port and do "killall -9 XFree86" - after X restarts it works fine.
It might be X rather than the nVidia drivers, but I doubt it (my old computer, also with X 4, works fine with ye olde Voodoo3).
Only if you are running the server, or have it mapped to a network drive or in Network Neighbourhood (Windows) or mounted on your filesystem somewhere (Unixes).
If you click either of these, you will be downloading the copy of the file on *your* computer. Not anyone else's, certainly not a remote server's.
The only way you can access files located on another computer by file:// URLs is if the other computer has shared them by Windows Networking or Samba or NFS or something, and you've mounted them on your local filesystem. (Actually that's not quite true, Windows will interpret file://mywinntbox/sharename/file as the NetBIOS UNC path \\mywinntbox\sharename\file, so if the file is shared with Windows Networking or Samba you don't need to have mounted the share as a network drive)
On the other hand, typing a URL rather than following a link will retrieve any resource that URL points to, so yes, you can get a certain amount of security-through-obscurity like that. (example: many web servers have/serverstatus or/admin which isn't usually linked to, although those should be password-protected too)
AFAIK, there is precisely one exception to the rule where nobody is directly allocated a second-level domain - nic.uk is the domain of the UK domain registrar, who can award themselves any.uk domain they feel like:-) and chose "nic" (presumably if ICANN gave.uk to some other company, they'd point www.nic.uk to the new one's web server).
However, there are quite a few fairly obscure UK 2LDs other than the well-known.co.uk,.org.uk,.net.uk,.ac.uk (academic, like.edu) and.gov.uk.
Examples include.nhs.uk (National Health Service - public-funded hospitals and so on),.police.uk (police forces, like cambs.police.uk for Cambridgeshire), and.ltd.uk and.plc.uk (only available to registered UK companies, and they have to use the closest domain-compatible version of their official name - e.g. FooSoft UK Limited would have to use foosoft-uk.ltd.uk and Whatevername Industries PLC would have to use whatevername-industries.plc.uk)
Slashdot Mirror
OK, I never said it was perfect :-)
Sun did make the mistake of assuming everyone had their own domain, but I think they were assuming only companies and similarly sized organizations would be writing anything much bigger than an applet. You could always have used nz.email.lpg_name for your previous project - it doesn't really matter, as long as nobody else would want that package name.
This scheme is intended for library-type classes; I suspect the "executable" for many Java projects is in the "null package" (i.e. the project's root directory).
When I said "most significant" I meant it in the sense of "most significant bit in a byte". If I write two related classes called SuperWidget and UberWidget which do stuff, and you write classes with the same names which do different stuff, it makes sense for my widgets to be in the same place in the directory tree, and your widgets to be somewhere else in the directory tree, rather than putting our (possibly unrelated) SuperWidgets in the same folder because they happen to share a name.
I agree, for small-scale or amateur developers, this scheme isn't ideal - but it really doesn't matter what the exact package name is, as long as nobody else will want it. It's like XML namespaces - an XML namespace is represented by a URI (usually an http:// one), but there's no need for there to actually be anything at that address (some people put the DTD or XML schema there, but the W3C actually discourage that because they don't want people to get used to the idea that namespaces are anything except an abstract identifier). They're only using domains and URIs at all in order to borrow an existing unique identifier, instead of having to implement some sort of separate registering system.
The global namespace thing is just that each package's full name includes an identifier for the server it's on.
9 58 559
Here's an analogy:
http://slashdot.org/comments.pl?sid=27540&cid=2
There's a uniquely identified item in a global namespace. I might use a mirror (web proxy) or an untrustworthy source (r00ted web proxy) to access it; that's my problem.
/comments.pl?sid=27540&cid=2958559
There's an item in a local namespace. Exactly the same as the one above, but
you need to know the context (i.e. that we're on Slashdot) before you can actually retrieve it.
If you decide to download, for example, a snapshot of gcc@gnu.org--gcc--3.0 (as far as I can see, that's roughly what gcc 3's unique ID might look like if the authors used arch) from l33t.h4x0r.net instead of ftp.gnu.org, that's your security problem, not the identifier's. If you don't think the site you're connecting to is the canonical source, don't use it - just like you'd download gcc-3.0.tar.gz (or whatever the distribution filename is) from a reputable mirror rather than l33t.h4x0r.net at the moment.
The canonical package name for your widgets would be nz.net.neural.(anything you like here)
.jar files, which have their own internal directory structure (they're slightly modified zip files).
If you own multiple domains (subdomains, or not), you pick one or more to use. The most sensible strategy would be to pick the one you were most likely to keep. Whether it corresponds to a real web page, or server, or whatever really doesn't matter - all that matters is that you control the neural.net.nz domain, and you don't use the same package name for different things as anyone else at that domain.
You do use directories for package name components - the class file for nz.net.neural.widgets.Widget (the convention is for class names to have initial caps) should go in nz/net/neural/widgets/Widget.class (replace / with your OS's directory separator if you don't use Unix). You often don't see this because classes are in
The domain has to be written backwards to put the most significant part first (otherwise neural.net and neural.net.nz would have overlapping namespaces, even though they might be owned by different people).
When did your Windows box last upgrade its kernel, then?
:-) From what I've gathered about Red Hat, Windows has much the same approach as a stock install of RH, in fact - you get a standard, intensively-tested kernel which isn't exactly on the cutting edge, but is stable.
So Windows Update consists of DLL upgrades (or shared library upgrades, to use the Linux terminology) - a lot like, say, an APT update on Debian, or whatever the equivalent's called on Red Hat.
If you want to be messing with custom-compiled kernels, that's your choice, and it's not as trivial as updating system libraries and programs. But then it's not a choice Windows even gives you, so I can't say I'm complaining
One thing Microsoft's strategy does have in its favour is that kernel-specific device drivers aren't a problem when all the copies of, say, Win2k have the same kernel. However, don't some Linux distros distribute packages of kernel modules compiled for their standard kernel (like the modified 2.4.9 kernel Red Hat use)?
<meta name="GENERATOR" content="Mozilla/4.72 [en]C-CCK-MCD VOYN-472 (Win98; I) [Netscape]">
Say no more.
That's exactly what /usr/local is for - locally compiled software. On most GNU and GNUish software the author sets it up to install to /usr/local by default, but you can do ./configure --prefix=/usr if you're building a distro package.
I don't know what other distros are like about this (I've only ever used Mandrake and Debian, and I didn't get experienced enough with Mandrake to know any of the internals), but Debian source packages come in two parts - a tarfile of original, unmodified source, plus a .diff.gz file containing the changes ("Debianizations") the Debian package maintaniner made to make it fit in with Debian conventions (moving all documentation to /usr/share/doc/name-of-the-package, for instance). If the original author's makefile or other code doesn't conform to Debian conventions, the maintainer will change it so it does.
For a program like you describe where (presumably) /usr/local/bin is hard-coded somewhere, the diff would include replacing that with /usr/bin - you, as an "upstream" developer, can probably make this easier by defining PREFIX to /usr/local and always referring to "$PREFIX/bin" and so on.
The things behind dpkg and all of that sure, they are nice, but how are they all that really different from RPM? I agree that I may not be totally clued in about all aspects of apt, dpkg and all of the other stuff Debian uses, but to say that it's so tied to a packaging format??? Debian is no more tied to it then Redhat is tied to RPM. If they are, then the Debian project made a major mistake!
Debian's primary advantage is that its packages are consistent. When you select a package, it comes with a very specific set of dependencies. That, and the fact that Debian is the source of the vast majority of .deb packages (especially the sort of "core" packages likely to come up in a dependency), mean they can get away with a lot of splitting up components, compiling platform-independent bits once and binaries many times (Debian isn't just an i386 distro like many others), shared libraries that get shared properly, and all sorts of other consistency stuff. Where a release-focused distro might concentrate on making all the packages in version 9.2 (say) work together nicely and consistently, Debian does a pretty good job of keeping all the packages consistent, all the time. (OK, so it doesn't always work 100% on 'unstable', but if you wanted stability you wouldn't be running that version anyway).
Maybe now is the time for Debian to actually form a company or form a different way of making the decisions instead of democracy.
The democracy/meritocracy structure is exactly what Debian is about - it's run by a non-profit org. called Software in the Public Interest. They don't exist to make a profit, or to make themselves popular - they exist to make a distribution of good software freely available.
Maybe they need to modify the DFSG to be more lienient?
And that would help how?
If it's free (free as in DFSG), it goes in 'main' and goes on the official CDs. If it's free as in DFSG but requires non-free software (like a GPLed "helper" application for PGP for instance), it goes in 'contrib'. If it's not free-as-in-DFSG (this includes shareware and closed-source freeware), it goes in 'non-free' - contrib and non-free are easily available over ftp or as the last CD in the set. If it's not in non-free, either nobody's tried packaging it, or Debian aren't sure if it would even be legal for them to distribute it.
Again, things like the DFSG are why Debian exists.
I don't know, but there has to be someone who is going to draw a line in the sand and get the volunteers in action and get Woody released (and with a 2.4 kernel as well...).
It looks like it'll be installed with a 2.2 kernel by default, and a 2.4 kernel as an option (fairly easy to install thanks to the magic of apt/dpkg - install, reboot, Lilo offers you a choice of kernel). Woody already includes precompiled 2.4.17 kernels for 386, Pentium, 686 (PPro/P2/P3/Celeron), K6 and K7 (Athlon/Duron), and that's just the i386 builds.
OTOH, if Debian was a company, where would all those volunteers be? Probably starting their own distro or doing Linux from Scratch.
PHB's and Joe Sixpacks like hearing and seeing commercials like Mandrake Linux is compliant (it isn't but) with the LSB which means no matter where you buy or download your software it will work!
This is why volunteer efforts like Debian have a niche. They don't like hearing and seeing commercials, they like giving people a stable OS. They're not trying to make a profit, which is why they can get away with doing what they feel is The Right Thing.
while a distro that provided an installer script would be fine. Indeed, Debian does precisely that as part of its 'contrib' archive. (As do other distros, probably)
Security, when you have to log in as root?
OK, so it's his mailing list, and his e-mail, so he has a right to do this. Even so:
Should self-proclaimed free software advocates really lower themselves to Microsoft's level? I'm sure this sort of elitism hurts the free software cause more than it helps it.
"I can't read your badly-formatted documents, so I'll deliberately stop your badly-written mail client reading mine. :P"
Sound childish to anyone?
Isn't the way to promote free software to be better than the competition? Microsoft can get away with assuming everyone uses Windows + Outlook Express, because we already know they're arrogant. If you claim to represent free software, being that arrogant risks having free software labelled as arrogant and elitist too.
Yes, RMS was right to be critical of using Word docs as attachments (I don't always agree with RMS's opinions, my opinion tends to be somewhere between open source and FSF). Critical is fine. If asked, most people on /. would be critical of Microsoft Out-of-order. But I can't really agree with deliberately breaking it.
Yes, IE 4 supports at least the most commonly used bits of CSS. NS4 is much worse (bizarre paragraph spacings and overlapping text and images were the main problems my site had on it).
Luckily there's a related Netscape bug. If you import your stylesheets with
@import url(/styles/mystyle.css);
instead of using the tag, Netscape 4 will completely ignore them, meaning NS4 users get a plain black-on-grey unstyled site - not too pretty, but at least they can read it without bits of it overlapping.
Windows limits users' capabilities by user interface as well as by permissions. If you have sysadmin-type priviledges (which you always do on Win9x) you can probably do raw hardware access (I'm thinking hard disks here), but not easily. You can format disk partitions, but you'll always be asked to confirm it. You can only make and delete partitions by rebooting to DOS mode (at least on Win9x, I'm not sure about NT-based versions like 2k and XP).
/dev/zero > /dev/hda", which will cheerfully fill your partition tables and disk with zero bytes). This is great if you know what you're doing, which is why Linux users are encouraged to always use a mere mortal userID, and only switch to root when they know exactly what they're doing (well, what they're about to do, anyway).
In Linux (as far as I can see) the user is assumed to be right, as long as they have sufficient priviledges. (Think "cat
This is why running as an admin-level user on Windows is pretty common, while running as root on Linux is, er, frowned upon.
For example: I know someone who works from home and occasionally goes in to the office, and uses the same (company-owned) laptop in both places. Because he needs to be able to alter network settings to switch between work LAN and home cable, the company's sysadmin (i.e. the guy with the laptop's Administrator password) has added admin-level access priviledges for that laptop to the login he usually works in.
If that was a Linux box, I assume the sysadmin would add the user to the list of sudo users, or even just give him the root password; either way, he'd have to make a conscious decision "I need super-user access for this; I'd better be careful".
Pseudorandom number generators such as the one used by rand() are predictable. AFAIK, /dev/random is extremely unpredictable because it uses hard-to-reproduce external data as an extra source of randomness - IIRC, it uses the precise (nearest millisecond?) timing of interrupts from the network cards, keyboard, mouse, etc. Not something you'll be able to replicate easily!
1) Notice it says "Can the developer of a program who distributed it under the GPL later license it to another party for exclusive use?" - there's nothing wrong with licensing it to another party for non-exclusive use. Look at the license on Cygnus Solutions' Cygwin DLL, which provides glue code to emulate a sensible Unix environment on Windows. Cygnus (now owned by Red Hat) dual-licenses it - the first is GPL (with extra permission to link to any OSI-approved open source software) and is free (as in free beer), the second is closed (but you can legally link it to a non-free app) and costs money. Borland are doing this too, with Kylix - the free (beer) downloadable version comes with GPLed runtimes, so you can only legally compile GPLed software on it, but the commercial version comes with the same libraries under Borland's usual "once you've paid us, you can link it to anything" licensing. 2) I suspect this is dodgy ground legally, and depends where X and Y live. If Y sends a few lines of diffs, they're probably not significant enough to be covered by copyright (your laws may vary). If Y explicitly puts their contributions in the public domain, X can claim it as their own and do what they like with it (that's what public domain means). If Y explicitly signs over copyright on their contributions (like the FSF requires on contributions to GNU code), X now owns it and can do what they like. Otherwise, X could probably argue that Y had implicitly given them permission to use their contributions... I think your quote is probably right in this case, but it seems like dodgy legal ground, and IANAL.
Focus-stealing when a new window appears depends on your window manager. The one built into KDE (does it have a name, or is it just "KDE"?) probably always steals focus. Sawfish can if you want it to (it does by default, I think; Customize->Focus to change its behaviour). Those are the only ones I've used a lot, so YMMV with others.
Are we benefiting MS by using Hotmail, though? My guess is the millions of disposable spam accounts they host soak up bandwidth and/or disk space without Microsoft actually gaining anything from the whole thing (at least on balance).
Yeah, OK, I meant it's not called the stable distribution :-) and there aren't official ISOs. I hadn't realised there were Sid ISOs too, though...
(I'm using unstable at the moment, the only serious instability I've noticed is the nvidia X driver which Debian have nothing to do with)
2.2r4 is the latest official CD-image. It's potato release 4, not woody (potato = 2.2, woody = 3.0). When Debian say "the stable version", that's what they mean... their aim is to get each release to all work together nicely, and they don't release that often. woody contains XFree86 4, gcc 3 and other newer goodies, but it hasn't become the stable distribution (it's still classified as testing) so there are no official ISOs.
They don't (or try not to) do ISOs as such, but the recommended way to make a Debian ISO is to download the packages to compile a "pseudo-image", then use rsync to turn that into the real ISO. They've done a convenient "pseudo-image kit" for Unixes and Windows (AFAIK, it may do more platforms) which consists of executables (probably source for the Unix version, I don't know since I used the Windows one), a readme, and pointers to where to get package mirrors, rsync mirrors and package lists from.
Using the pseudo-image kit (even on Windows) is at least as easy as actually installing Linux (read instructions, follow instructions - I get the impression Debian is aimed at people who can read instructions, so that suits them fine).
The resulting CD is bootable and works fine (and once you've finished with it, its bootability, built-in root FS and utilities make it a nice rescue disk).
Basically, with the GPL under copyright, you can say "this software is licensed under the GNU GPL" and it's unambiguous what the terms are.
This isn't necessary for, say, BSD-style licenses (BSD, ZLib, LibPNG) because they're simpler and shorter - it's reasonable to include the entire BSD license in each file of your source, so you don't say "this is BSD-licensed", you say "you may do this, this and this but not this". "BSD license" is just a convenient shorthand for describing things - but from a legal point of view, the license consists of a couple of paragraphs embedded in each source file.
However, it's obviously not reasonable to include the whole GPL in the same way. The GPL is long (20K?) because of copyleft - it's less permissive than the BSD license, so it can't just grant blanket permissions like the BSD license does (although an abbreviated GPL without the preamble/manifesto would be nice, since they're not really part of the license as such).
If the GPL was free (in the FSF sense of the word) or open source, you'd get people redefining what it meant, and much confusion would ensue. ("Our software is licensed under the GPL." "No it isn't, ours is the real GPL!")
Network cables and some business phone jacks, however, are physically compatible and can be indistiguishable. I have made that mistake, and you would as well. Blame industry for it: choosing a phone cable standard for Ethernet was NOT a smart move.
:-)
I can agree with that. Each student room in the college I'm at has a double socket on the wall - one RJ45 network connection, and one physically identical phone socket (which you need an adaptor to plug a normal phone into). All we were told was that the network connection was *usually* the one on the left
It's interesting that in The Feeling of Power, the guy who rediscovers arithmetic does so by reverse-engineering old obsolete computers and slowly working out how they do it :-)
They're not perfect. I don't get crashes that often, but occasionally (once every couple of days, usually while in a game but sometimes just randomly) X freezes and won't even respond to keyboard input. Luckily I have a palmtop (an old Psion 3c) and a serial cable, so I can still log in on a serial port and do "killall -9 XFree86" - after X restarts it works fine.
It might be X rather than the nVidia drivers, but I doubt it (my old computer, also with X 4, works fine with ye olde Voodoo3).
Only if you are running the server, or have it mapped to a network drive or in Network Neighbourhood (Windows) or mounted on your filesystem somewhere (Unixes).
/serverstatus or /admin which isn't usually linked to, although those should be password-protected too)
file:// links to files on your local filesystem.
Examples:
file:///c:/Windows/Notepad.exe - a well-known executable on a Windows 9x box
file:///bin/bash - a well-known executable on a Linux box
If you click either of these, you will be downloading the copy of the file on *your* computer. Not anyone else's, certainly not a remote server's.
The only way you can access files located on another computer by file:// URLs is if the other computer has shared them by Windows Networking or Samba or NFS or something, and you've mounted them on your local filesystem. (Actually that's not quite true, Windows will interpret file://mywinntbox/sharename/file as the NetBIOS UNC path \\mywinntbox\sharename\file, so if the file is shared with Windows Networking or Samba you don't need to have mounted the share as a network drive)
On the other hand, typing a URL rather than following a link will retrieve any resource that URL points to, so yes, you can get a certain amount of security-through-obscurity like that. (example: many web servers have
AFAIK, there is precisely one exception to the rule where nobody is directly allocated a second-level domain - nic.uk is the domain of the UK domain registrar, who can award themselves any .uk domain they feel like :-) and chose "nic" (presumably if ICANN gave .uk to some other company, they'd point www.nic.uk to the new one's web server).
However, there are quite a few fairly obscure UK 2LDs other than the well-known .co.uk, .org.uk, .net.uk, .ac.uk (academic, like .edu) and .gov.uk.
Examples include .nhs.uk (National Health Service - public-funded hospitals and so on), .police.uk (police forces, like cambs.police.uk for Cambridgeshire), and .ltd.uk and .plc.uk (only available to registered UK companies, and they have to use the closest domain-compatible version of their official name - e.g. FooSoft UK Limited would have to use foosoft-uk.ltd.uk and Whatevername Industries PLC would have to use whatevername-industries.plc.uk)