That's precisely what apt does since the end of the last century.
Don't worry, I'm well aware of APT (being an avid Debian user). I meant the case of J. Random User double-clicking in a file browser window on the icon for a random RPM or.deb file s/he happened to download from somewhere (not via APT), since InstallShield fans apparently want to install software this way.
RPM, deb, ebuilds, tar.bz2, tar.gz, all are to complicated for the normal user.
I have trouble understanding why comments like this keep coming up. RPM, deb, tar.gz, and so on aren't installation programs. They are package formats.
Even leaving aside the whole question of whether an integrated package manager like Synaptic, KPackage or RHN is easier to use (it certainly is!) than for the user to download software manually from all over the place...
Users don't need to have a clue about the actual file format of these things; they just need to be able to double-click on one of these files in (say) a Nautilus window, causing the underlying package manager to pop up a "Root password?" dialog box, then automatically install the package. What could be simpler than that? From the user's point of view, how is this any different from double-clicking on the icon for a Windows installer program generated with InstallShield?
Admittedly we may not quite be at that point yet -- if the RPM/deb has unmet dependencies, for instance, then the package manager should automatically download and install those as well when the file is double-clicked -- but we're getting there fast. And Windows-style executable installers, for reasons of consistency, are NOT the way to go.
Slightly off-topic: anyone ever try binary editing of a deb file to put "#!/bin/dpkg -i" at the beginning, or "#!/bin/rpm -i" at the beginning of an RPM, and chmod a+x'ing it? Does it work?
So: can't they just block modem calls to these countries? It would still stop people sending faxes, but would not affect the vast majority of legitimate calls...
That's why your suggestion won't fly; a hell of a lot of businesses and other organizations still rely on international faxes. E.g., my research experiment is based in Italy. The ability to fax signed paperwork from here (the US) to there, and back, is crucial.
Description: ELF prelinking utility to speed up dynamic linking The prelink package contains a utility which modifies ELF shared libraries and executables, so that far fewer relocations need to be resolved at runtime and thus programs come up faster.
Used to make big C++ programs like OpenOffice, KDE, etc. load faster, but of course it has to modify the binaries to do so. I suppose this would break binary diffs against them.
Go to the "Web Browser" tab in the same dialog box, check the "custom web browser" button and enter "mozilla-firefox '%s'" into the text entry, I suppose. Haven't tried this personally though.
It depends on if you're running Windows or something else. On Windows, it behaves as you'd expect. On Linux, I'm still doing the old copy-location, paste-location trick.:(
If you're running Gnome 2.6, go to Applications -> Desktop Prefs -> Advanced -> Preferred Applications, and select the Mail Reader tab. Then check the Custom Mail Reader radio button and enter
mozilla-thunderbird -compose '%s'
into the text entry box. It works for me, at least (running Debian unstable). Too bad this isn't automatic.
It is a fact that my iBook is better made than the Toshiba I had before and it is clearly holding up to daily use better than the Toshiba.
I love my iBook, but the hard disk on it died with a horrible grinding sound recently, after only a year and a half. Ironically enough, the hard disk was in fact made by Toshiba. Perhaps tempting fate, I replaced it (what a bitch!) with another Toshiba (upgraded from 20 to 40 Gb), although I've been making twice-weekly backups since then.
Oh, and OS X is much nicer than Linux on a laptop. That is a fact and I have been a Linux user since 1994 (before that I used SunOS).
Here I have to disagree:-) I'm using Debian GNU/Linux because of the utter convenience of upgrading and software installation, and the superb development environment for free software it provides. (Yes, I know about Apple's IDE, but I don't want to learn Objective C and the GUI isn't at all portable to other UNIXes.)
Plus, Linux and the vast quantity of software packaged in Debian are free. To be able to use the latest Fink packages, OTOH, I would either have to spend forever compiling them, or upgrade OS X to 10.3, which is a not-insignificant sum of money for a grad student. I don't think I've even booted into OS X since re-installing, except from inside MOL a few times.
Assuming you [the seller] give them multiple avenues to contact you, then they [the buyer] simply aren't that interested if they only send one email and drop it after that.
Either that, or the prospective buyer sent an email to a number of different companies in order to see who responded fastest / with the best quote / most reassuringly. And by missing that first email, you lost automatically.
in true slashdot bashing style. Haha!
i kinda wonder if linux offends accidentally or not tho.
Actually, the well-known Debian developer Herbert Xu
resigned recently due to
being offended by project members discussing (in the context of the new Debian installer) whether or not to refer to Taiwan as "Taiwan, Province of China" as it is officially designated in ISO 3166.
So this sort of problem is certainly not restricted to commercial or proprietary software.
I switched from Pine to Thunderbird a few weeks ago; here are the most important things I miss:
The ability to mark certain lines not to be automatically wrapped when composing a text email. This is important when reporting error messages from compiler output, etc.
The ability to include a text file inline in the email message (NOT as an attachment) while composing it. Useful when including config files, quotes from more than one email at once, etc.
Another feature which would be nice to have (but not nearly as important to me) is support for mbox folders in subdirectories of the top-level mail folder.
Anyone know whether it's possible to do any of the above in Thunderbird? If not, what's the best way to make the feature request?
Will the new Thunderbird icons be made available under the same license as the Firefox icons? There unfortunately seem to be some issues with using them in the packages provided by various Linux distributions; please see this thread for details:
The PowerPC kernel, for instance, exists as its own set of patches to the main kernel tree.
<pedantic>
Actually Linus switched to using an Apple laptop sometime fairly recently, so as of about 2.6.3, the PowerPC kernel is identical to the vanilla kernel tree.
The GPL applies to the entire work, but ONLY the entire work. It does not apply to stuff that the licensor did not create. Specifically, a KDE program under the GPL does not virally infect the X11 libraries. They are SEPARATE works.
Of course a library doesn't become GPL-ized just because a GPL program links against it. But the "combined work" consisting of the program + shared libraries inherits ALL of the licenses of its components. This is at least the interpretation of GNU, as stated in the preamble to the LGPL:
When a program is linked with a library, whether statically or using a shared library, the combination of the two is legally speaking a combined work, a derivative of the original library. The ordinary General Public License therefore permits such linking only if the entire combination fits its criteria of freedom.
As you point out, the existence of several libraries with the same API but under different licenses weakens this claim. But I suspect most Linux distributions intend to respect it in order to take the most conservative path (legally speaking), so as not to become a test case for this claim in court.
Gnome runs on Solaris. Solaris ships with a GPL-incompatible proprietary Openwindows X11 implementation. Yet Sun also ships with GPL applications (like Gnome) that link to it.
In this case Sun is probably taking advantage of the "system libraries" exception in GPL section 3:
However, as a special exception, the source code distributed need not include anything that is normally distributed (in either source or binary form) with the major components (compiler, kernel, and so on) of the operating system on which the executable runs, unless that component itself accompanies the executable.
I admit that I don't see how they are getting around the "unless that component itself accompanies the executable" clause. But in any case, your statement that the GPL applies only to an executable, and not to the combination of executable + shared libraries, is at best undecided.
Now, the question I have to ask is this: How can the GPL apply to "an entire work"? If I write something, only the part that I'm writing (or modify) should be under the GPL.
First, the "entire work" consisting of program + libraries is not under only the GPL, but inherits the licenses of ALL its components. If some of the licenses are incompatible (like the GPL and new XFree86 licenses), the work can't be distributed at all.
Believe it or not, there's a good reason for this "entire work" consideration. Suppose that the GPL did not treat the combination "program + shared libraries" as a single work. Suppose further that Evil Corporation Inc. (ECI) wants to use code from a GPL-licensed program, let's say Frozen Bubble, for its proprietary product. All that ECI has to do is rewrite Frozen Bubble to be a shared library, and then it can link its proprietary product to the shared-library version of Frozen Bubble (which, to comply with the legalities, ECI still makes available under GPL).
So if the GPL considered programs as separate entities from the shared libraries they are linked against, it would reduce the GPL to be equivalent to the LGPL (which DOES permit linking proprietary programs against LGPL libraries). The LGPL is a fine license, but presumably the authors of Frozen Bubble didn't want their code used AT ALL by a proprietary product, so they used the GPL instead.
So the entire argument is about being forced to give credit where credit is due?
Only if you oversimplify. Here's an executive summary that I hope isn't oversimplified:
XFree86 new license requires that people who redistribute modified versions of XFree86 include specified people in the credits of their programs or documentation.
GPL does not require that.
GPL specifically says that it is NOT COMPATIBLE with licenses that put restrictions on the user that the GPL does not.
GPL applies to an entire work, e.g. a GNOME or KDE program plus all the libraries it's linked against.
Therefore, distributions have NO LEGAL RIGHT to distribute GPL-licensed programs linked against X libraries when the libraries are licensed under the new XFree86 license.
There is nothing morally wrong with the new XFree86 license, and it is probably even "free software" or "open source" by most definitions. The problem is that it is NOT COMPATIBLE with the majority of existing free software. If the situation were reversed, most free software were licensed under what we call the new XFree86 license, and suddenly XFree86 wanted to relicense all of its code under a radical new license they called the GPL, the complaints would be exactly the same: the new license would be incompatible with the vast majority of existing open source code.
What law? The GPL isn't a law, the BSD license isn't a law. They're just licenses.
Yes, but with the current state of copyright laws, those licenses are the only things that give you legal permission to use or distribute a program. Said permission does not exist by default. So if you violate a license, you no longer have legal permission to use a program, and by definition you are then breaking copyright law.
Yeah, but Debian is no longer supporting nonfree software with their next (sarge) release. Doesn't that put XFree86 back on the Outs with Debian?
Sorry, but your first sentence is completely wrong. The amendment to remove support for non-free in Debian didn't even get a majority, let alone the 3:1 majority it would have required to win. See these links: Debian Planet, official Debian vote results.
It is true that no one on Debian's X Strike Force has intentions to package any of the XFree86 releases that the new license is applied to. But since Debian is an entirely volunteer project, conceivably a developer who really wanted to embark on this thankless task could do so, although the packages would probably have
to go into non-free.
For physics (especially particle physics),
Statistical Data Analysis by Glen Cowan is very good at the grad student level. The book has its own home
page here.
The function e^x and a constant function were wandering around town towards sunset. The constant function told his friend, "I'm going home now. At night the differentials come out, and if they operate on me, I'm a goner."
"Suit yourself," replied e^x. "They can't do anything to me."
Sure enough, after another half hour a differential leaped out of an alley. Full of bravado, e^x introduced himself, "Hi! I'm e^x."
It was introduced in 2.4.23-pre7, disguised as "Add TASK_SIZE check to do_brk()" in the changelog.
If you aren't averse to compiling your own kernel, the fix is a really easy two line patch. Just add the following to mm/mmap.c at line 1047 (immediately after "if (!len) return addr;")
I package [9] a large set of open source software programs and libraries [10] developed at CERN for the Debian GNU/Linux project [11], one of the most popular Linux distributions.
That should read more like "I package a large set of open source programs and libraries, developed by CERN, for the Debian GNU/Linux project...". Obviously I didn't want to imply that CERN wrote this software specifically for Debian. It was 2 am when I wrote the above...
Slashdot Mirror
That's precisely what apt does since the end of the last century.
Don't worry, I'm well aware of APT (being an avid Debian user). I meant the case of J. Random User double-clicking in a file browser window on the icon for a random RPM or .deb file s/he happened to download from somewhere (not via APT), since InstallShield fans apparently want to install software this way.
RPM, deb, ebuilds, tar.bz2, tar.gz, all are to complicated for the normal user.
I have trouble understanding why comments like this keep coming up. RPM, deb, tar.gz, and so on aren't installation programs. They are package formats.
Even leaving aside the whole question of whether an integrated package manager like Synaptic, KPackage or RHN is easier to use (it certainly is!) than for the user to download software manually from all over the place... Users don't need to have a clue about the actual file format of these things; they just need to be able to double-click on one of these files in (say) a Nautilus window, causing the underlying package manager to pop up a "Root password?" dialog box, then automatically install the package. What could be simpler than that? From the user's point of view, how is this any different from double-clicking on the icon for a Windows installer program generated with InstallShield?
Admittedly we may not quite be at that point yet -- if the RPM/deb has unmet dependencies, for instance, then the package manager should automatically download and install those as well when the file is double-clicked -- but we're getting there fast. And Windows-style executable installers, for reasons of consistency, are NOT the way to go.
Slightly off-topic: anyone ever try binary editing of a deb file to put "#!/bin/dpkg -i" at the beginning, or "#!/bin/rpm -i" at the beginning of an RPM, and chmod a+x'ing it? Does it work?
I hope some FBI employee had a good laugh over this one.
Hmm, isn't filing a false police report a felony? I hope they will do more than have a laugh...
So: can't they just block modem calls to these countries? It would still stop people sending faxes, but would not affect the vast majority of legitimate calls...
That's why your suggestion won't fly; a hell of a lot of businesses and other organizations still rely on international faxes. E.g., my research experiment is based in Italy. The ability to fax signed paperwork from here (the US) to there, and back, is crucial.
Here's another potential problem:
Used to make big C++ programs like OpenOffice, KDE, etc. load faster, but of course it has to modify the binaries to do so. I suppose this would break binary diffs against them.
Go to the "Web Browser" tab in the same dialog box, check the "custom web browser" button and enter "mozilla-firefox '%s'" into the text entry, I suppose. Haven't tried this personally though.
It depends on if you're running Windows or something else. On Windows, it behaves as you'd expect. On Linux, I'm still doing the old copy-location, paste-location trick. :(
If you're running Gnome 2.6, go to Applications -> Desktop Prefs -> Advanced -> Preferred Applications, and select the Mail Reader tab. Then check the Custom Mail Reader radio button and enter
into the text entry box. It works for me, at least (running Debian unstable). Too bad this isn't automatic.
It is a fact that my iBook is better made than the Toshiba I had before and it is clearly holding up to daily use better than the Toshiba.
I love my iBook, but the hard disk on it died with a horrible grinding sound recently, after only a year and a half. Ironically enough, the hard disk was in fact made by Toshiba. Perhaps tempting fate, I replaced it (what a bitch!) with another Toshiba (upgraded from 20 to 40 Gb), although I've been making twice-weekly backups since then.
Oh, and OS X is much nicer than Linux on a laptop. That is a fact and I have been a Linux user since 1994 (before that I used SunOS).
Here I have to disagree :-) I'm using Debian GNU/Linux because of the utter convenience of upgrading and software installation, and the superb development environment for free software it provides. (Yes, I know about Apple's IDE, but I don't want to learn Objective C and the GUI isn't at all portable to other UNIXes.)
Plus, Linux and the vast quantity of software packaged in Debian are free. To be able to use the latest Fink packages, OTOH, I would either have to spend forever compiling them, or upgrade OS X to 10.3, which is a not-insignificant sum of money for a grad student. I don't think I've even booted into OS X since re-installing, except from inside MOL a few times.
Assuming you [the seller] give them multiple avenues to contact you, then they [the buyer] simply aren't that interested if they only send one email and drop it after that.
Either that, or the prospective buyer sent an email to a number of different companies in order to see who responded fastest / with the best quote / most reassuringly. And by missing that first email, you lost automatically.
in true slashdot bashing style. Haha!
i kinda wonder if linux offends accidentally or not tho.
Actually, the well-known Debian developer Herbert Xu resigned recently due to being offended by project members discussing (in the context of the new Debian installer) whether or not to refer to Taiwan as "Taiwan, Province of China" as it is officially designated in ISO 3166.
So this sort of problem is certainly not restricted to commercial or proprietary software.
I switched from Pine to Thunderbird a few weeks ago; here are the most important things I miss:
Another feature which would be nice to have (but not nearly as important to me) is support for mbox folders in subdirectories of the top-level mail folder.
Anyone know whether it's possible to do any of the above in Thunderbird? If not, what's the best way to make the feature request?
Will the new Thunderbird icons be made available under the same license as the Firefox icons? There unfortunately seem to be some issues with using them in the packages provided by various Linux distributions; please see this thread for details:
Debian Legal thread on Firefox trademark issuesAnd we know how much faith creationists place in radioactive dating methods... :-)
Um... because it has a lot of GNU apps, all packaged together, the way you seemed to want?
The PowerPC kernel, for instance, exists as its own set of patches to the main kernel tree.
<pedantic>Actually Linus switched to using an Apple laptop sometime fairly recently, so as of about 2.6.3, the PowerPC kernel is identical to the vanilla kernel tree.
</pedantic>
The GPL applies to the entire work, but ONLY the entire work. It does not apply to stuff that the licensor did not create. Specifically, a KDE program under the GPL does not virally infect the X11 libraries. They are SEPARATE works.
Of course a library doesn't become GPL-ized just because a GPL program links against it. But the "combined work" consisting of the program + shared libraries inherits ALL of the licenses of its components. This is at least the interpretation of GNU, as stated in the preamble to the LGPL:
When a program is linked with a library, whether statically or using a shared library, the combination of the two is legally speaking a combined work, a derivative of the original library. The ordinary General Public License therefore permits such linking only if the entire combination fits its criteria of freedom.
As you point out, the existence of several libraries with the same API but under different licenses weakens this claim. But I suspect most Linux distributions intend to respect it in order to take the most conservative path (legally speaking), so as not to become a test case for this claim in court.
Gnome runs on Solaris. Solaris ships with a GPL-incompatible proprietary Openwindows X11 implementation. Yet Sun also ships with GPL applications (like Gnome) that link to it.
In this case Sun is probably taking advantage of the "system libraries" exception in GPL section 3:
However, as a special exception, the source code distributed need not include anything that is normally distributed (in either source or binary form) with the major components (compiler, kernel, and so on) of the operating system on which the executable runs, unless that component itself accompanies the executable.
I admit that I don't see how they are getting around the "unless that component itself accompanies the executable" clause. But in any case, your statement that the GPL applies only to an executable, and not to the combination of executable + shared libraries, is at best undecided.
Now, the question I have to ask is this: How can the GPL apply to "an entire work"? If I write something, only the part that I'm writing (or modify) should be under the GPL.
First, the "entire work" consisting of program + libraries is not under only the GPL, but inherits the licenses of ALL its components. If some of the licenses are incompatible (like the GPL and new XFree86 licenses), the work can't be distributed at all.
Believe it or not, there's a good reason for this "entire work" consideration. Suppose that the GPL did not treat the combination "program + shared libraries" as a single work. Suppose further that Evil Corporation Inc. (ECI) wants to use code from a GPL-licensed program, let's say Frozen Bubble, for its proprietary product. All that ECI has to do is rewrite Frozen Bubble to be a shared library, and then it can link its proprietary product to the shared-library version of Frozen Bubble (which, to comply with the legalities, ECI still makes available under GPL).
So if the GPL considered programs as separate entities from the shared libraries they are linked against, it would reduce the GPL to be equivalent to the LGPL (which DOES permit linking proprietary programs against LGPL libraries). The LGPL is a fine license, but presumably the authors of Frozen Bubble didn't want their code used AT ALL by a proprietary product, so they used the GPL instead.
So the entire argument is about being forced to give credit where credit is due?
Only if you oversimplify. Here's an executive summary that I hope isn't oversimplified:
There is nothing morally wrong with the new XFree86 license, and it is probably even "free software" or "open source" by most definitions. The problem is that it is NOT COMPATIBLE with the majority of existing free software. If the situation were reversed, most free software were licensed under what we call the new XFree86 license, and suddenly XFree86 wanted to relicense all of its code under a radical new license they called the GPL, the complaints would be exactly the same: the new license would be incompatible with the vast majority of existing open source code.
What law? The GPL isn't a law, the BSD license isn't a law. They're just licenses.
Yes, but with the current state of copyright laws, those licenses are the only things that give you legal permission to use or distribute a program. Said permission does not exist by default. So if you violate a license, you no longer have legal permission to use a program, and by definition you are then breaking copyright law.
Yeah, but Debian is no longer supporting nonfree software with their next (sarge) release. Doesn't that put XFree86 back on the Outs with Debian?
Sorry, but your first sentence is completely wrong. The amendment to remove support for non-free in Debian didn't even get a majority, let alone the 3:1 majority it would have required to win. See these links: Debian Planet, official Debian vote results.
It is true that no one on Debian's X Strike Force has intentions to package any of the XFree86 releases that the new license is applied to. But since Debian is an entirely volunteer project, conceivably a developer who really wanted to embark on this thankless task could do so, although the packages would probably have to go into non-free.
sigh, so much for moderating this discussion...
For physics (especially particle physics), Statistical Data Analysis by Glen Cowan is very good at the grad student level. The book has its own home page here.
We have moved most of our lab machines from Windows to OS X in the past few months...
Where did you find machines that can run both Windows and Mac OS X? I'd like one!
A variation:
The function e^x and a constant function were wandering around town towards sunset. The constant function told his friend, "I'm going home now. At night the differentials come out, and if they operate on me, I'm a goner."
"Suit yourself," replied e^x. "They can't do anything to me."
Sure enough, after another half hour a differential leaped out of an alley. Full of bravado, e^x introduced himself, "Hi! I'm e^x."
"Pleased to meet you, I'm d/dy."
It was introduced in 2.4.23-pre7, disguised as "Add TASK_SIZE check to do_brk()" in the changelog.
If you aren't averse to compiling your own kernel, the fix is a really easy two line patch. Just add the following to mm/mmap.c at line 1047 (immediately after "if (!len) return addr;")
if ((addr + len) > TASK_SIZE || (addr + len) < addr))return -EINVAL;
I'm enjoying the thrill of compiling patched kernels on two different machines as I write this. Thank goodness for Debian's make-kpkg.
That should read more like "I package a large set of open source programs and libraries, developed by CERN, for the Debian GNU/Linux project...". Obviously I didn't want to imply that CERN wrote this software specifically for Debian. It was 2 am when I wrote the above...