I remember the days. I came in right at the end of when JA Net could have been considered as something separate from the Internet; it still had its own protocols (I never knew much about them, but I do remember numeric addresses that were about 15 digits long with no separators), and its own conventions (they were starting to adopt internet naming, but a lot of stuff was backwards... eg uk.ac.warwick rather than warwick.ac.uk).
Didn't realise it went back as far as early 80s though. I knew of a few dial-in MUDs around 84. I think BT invested in one at some point, but that would have been later. 86, 87? Wonder how much revenue they got from that?
...that everything I did on company property or with company resources would be copyright AOL?
You have it lucky. For many, it doesn't matter whether or not company resources were used. Everything you do during the period in which you're working for them is up for grabs. At least AOL seem to make it clear where they stand.
There is one revenue model that, unfortunately, they won't be able to exploit, due to the nature of the GPL.
The GPL does not allow code to be incorporated into proprietary software; theoretically the code can be relicensed to allow this each time it is needed, but to do so would require the consent of everyone who ever contributed anything to the codebase (even just a single character typo-fix to a message, or whatever)!
Alternative licenses can be written that allow open source freedom just as much, but also allow such relicensing. The license of NASM was such a license, although many object to the fact that code submitted under this license effectively becomes the property of an arbitrary group of `authors'.
I'm working on a new licensing scheme that fixes this problem and which is somewhat better phrased than the NASM license. You can see my work-in-progress on my web-site, The Democratic Software Foundation.
I appreciate comments, although don't promise to reply quickly!
In response to Penrose's arguments, I have two points. He rests the majority of his case upon Gödel's theorem of incompleteness.
First: there is no evidence that Gödel's theories don't apply to human intelligence just as much as they do to the automated reasoning processes that they were originally targeted at debunking.
Second: Gödel's theories only apply to deterministic systems. The moment you start including random numbers in the equations, you cannot get a result anymore. Your automated reasoning system can break out of the 'rut' that it was in, and up to the next class of problems. Doesn't this sound exactly like the way humans approach mathematics? Ask any mathematician who has struggled on a hard problem for years - the solution comes in a moment of inspiration - just another word for a random thought unlocking the secrets of the problem.
So, thinking machines must be nondeterministic. Fine; we already know how to make circuits behave in a nondeterministic manner.
> If you want to get really fly you could store
> that file in a RAM disk.:)
Although there's really no point, cos it'll be read out of disk cache on every hit, if you've got the right page out there!
...for important problems in this sort of stuff come out of the commercial vendors as fast as they do from the FreeUNIXes
Sorry, but I have to disagree here. Case in point - a major bug was recently discovered in the internationalisation functions of the C libraries for most major Unices. Linux and Solaris were both affected, and several commonly used setuid programs turned out to contain root shell vulnerabilities. glibc was patched to fix the problem within days, but AFAIK to date (about 4 weeks later) there is still no fix on Solaris other than disabling the setuid facility of these programs.
To anyone relying on these facilities, this is a fairly large step up in TCO.
(Disclaimer: this isn't intended as a dig at Sun. They do a great job, but it is tough to compete with free software hackers in terms of speed getting fixes out...)
Re:Get your mainframe!! Mainframes here!!
on
User Mode Linux
·
· Score: 1
As a DSVR client, I must say that their service is
very good, except for a couple of problems:
- software that attempts to bind sockets to IPADDR_ANY can get confused by their system and have conflict with the other virtual servers on their machine. I have this trouble with mysql, which just about everyone runs there; either you have to turn off networking or use a non-standard port.
- you don't get a proper root account, and a lot of things don't work properly (eg admin can't access other users' files without relevant permissions).
A service using the user mode kernel would (imho) be significantly better. But then the DSVR approach is lower cost (in terms of CPU time) and sufficient for almost all uses.
Hmmm, never had the fortune to have all that add on stuff, but Parsec was fun. But then I got my old system working again a couple of years ago, and found that my skill had somewhat improved; just kept playing it until I was so totally bored.
The TI99/4 was, I understand, the first 16 bit home computer available to the mass market (predating the IBM PC by about 2 years). The 4A was a fairly simple upgrade.
A point that a lot of people have missed is that the 'common sense' argument you point out is currently held by the EU.
Under EU law, you can patent software, but only if it exhibits a 'novel technical effect'. My understanding of it is that this disallows any mathematically based or internal algorithmic patents (eg, the RSA patent or one that I heard about where somebody effectively had a patent on a binary search), but allows patents of new ways to make a system behave (eg BTs patent on something that approximates hyperlinks, as discussed recently) as long as nobody can show prior art.
Oooh! I just had another thought. They can furnish each potential visitor with a pantone card set and gamma correction software, and insist that they adjust their monitors until they get just the right shading. Yeah, that'll work.
Funnily enough, but I recently came across a web site (unfortunately I've lost the URL) that allows you to do exactly that. Your images are served through a cgi script on their server that gamma corrects them to account for the viewers monitor!
1. "this Licence" could be interpreted to mean the GPL. In that case, we are 100% compliant. Since "this Licence" applying to the NASM licence would be a contradiction, this is a reasonable interpretation
In what way does the latter interpretation cause a contradiction? I see none. Also, under the most rigorous English grammar, it is the only possible interpretation.
2. Even if "this Licence" does not refer to the GPL, that sentence would be deemed unenforcable (as it contradicts the earlier sentence), and the remainder of the agreement would remain in force
The NASM licence doesn't claim to be wholly compatible with the GPL.
X. In addition to what this Licence otherwise provides, the Software
may be distributed in such a way as to be compliant with the GNU
General Public Licence, as published by the Free Software Foundation,
Cambridge, MA, USA; version 2, or, at your option, any later version;
incorporated herein by reference. You must include a copy of this
Licence with such distribution. Furthermore, patches sent to the
authors for the purpose of inclusion in the official release version
are considered cleared for release under the full terms of this
Licence.
Essentially, this allows distribution of NASM in GPL code distributions, and just keeps the companies maintaining those distributions happy.
The NASM licence is no more restrictive than the GPL is. In fact it allows certain freedoms that the GPL doesn't.
To be honest, I don't put much faith in such metrics. They are based on rules of thumb and programming doctrine and cannot really get anywhere sensible.
Yes, it is a little difficult to understand the parser. It could perhaps have been better structured. But more than 2 or 3 people have worked with it, and most seem to be able to grasp it.
Try actually reading the code, rather than applying automated tests. You'll find it really isn't that bad.
And some of the 'bad programming practices' you mention elsewhere actually make the code easier to read; the global variables are used so that thousands of pointers don't have to be passed all over the place. Most of them don't change throughout the operation of the program and merely indicate configuration options. Others change infrequently, or represent values that are effectively 'read-only environment values' for most parts of the code and are only changed in very rare circumstances.
The Nasm licence now clearly states, "the Software may be distributed in such a way as to be compliant with the GNU General Public Licence". That is quite clear. I can download Nasm, and sell a comercial version based on it, so long as I stay compliant to the GPL, right?
As long as the code remains under the original licence, and is not simply relicenced under the GPL. That is the intention.
The Nasm license also states, "We may well allow you to incorporate it into commercial software too, but we'll probably demand some money for it". IANAL, but I would say that this is now unenforcable, as they have declared that I may redistribute under the terms of the GPL, right?
No. Just because we have declared that you can distribute where you could under the GPL, this doens't mean that if what you want to do is outside of the scope of the GPL (the intended meaning of 'incorporate into commercial software' above) this is in some way invalidated. Yes, if what you want to do is within the scope of the GPL we can't charge for it. We haven't charged SuSE, RedHat or any of the other companies charging for copies on their CDs.
Personally I think this is utter rubbish. So what if the GPL is more restrictive than the NASM licence? The NASM licence may not allow you to relicence under the GPL, but neither would the GPL allow you to relicence under any other!
In the end, what's the big deal about the GPL?
Just because its the licence that some big corps
(ie redhat et al) like, does that mean we have to kowtow to their judgement? OSS developers should be free to grant whatever freedoms they like to the people using their software. A few of the freedoms we wanted to grant with NASM (and yes, I don't mind admitting that profit was among the motivations!) are incompatible with the GPL.
Allowing nasm code to be distributed under the GPL would render this impossible, because then code submitted wouldn't be cleared for inclusion under the nasm licence, but only under the GPL.
This is not an 'imperfection in the wording' as you assume; it is fully intentional and designed to preserve the rights of potential users of NASM whose uses are not covered by the GPL. And that doesn't only stretch to proprietory systems. It allows binary only distributions (placing copies of NASM on magazine covers who don't want to be bothered by restrictions on carrying source code as well, etc). It allows linking NASM code with binary only but free projects (there are some out there; not that any use NASM to my knowledge, but you never know).
Indeed. GPL'd code cannot be incorporated into the NASM code without the permission of all of the authors of the GPL'd code. This is IMHO a weakness with the GPL, allowing rights over the code to be distributed among too many people. Truly free software should not make such restrictions. The NASM licence allows the NASM code to be used in situations where using GPL'd code would be impossible, because it allows a few people to make executive decisions concerning the copyright of the code.
As one of the originaly authors, we went through a stage shortly before we made our first public release of the nasm code where we wanted to make sure nasm was available to the widest range of people possible.
While the GPL is a fine licence, it suffers many shortcomings in our opinion. One of those is that it is riddled with free software political agenda. While we wanted nasm to be free, we also wanted it to be available for people who had non-free uses for it. Launching it simply under the GPL would have caused a mess; in cases where nasm's use is not allowed under the GPL (there have been several such instances), we would have to acquire the permission of everyone who ever contributed code in order to relicence outside of the scope of the GPL.
The NASM licence is designed to avoid this problem. It allows a small, executive group to take control of the rights of code submitted for inclusion in NASM, and to relicence them as they see fit.
In hindsight, there are other ways we could have gone that would have caused many problems.
But despite the contents of the e-mail that I received recently from Stallman, I am sure that the NASM licence does not break the GPL, and its inclusion within the licence means that other than where the rights of those who submit code for inclusion is restricted further, all other rights are extended beyond what they are in the GPL.
I'll post a few points elsewhere, where individual points have been made.
Slashdot Mirror
Didn't realise it went back as far as early 80s though. I knew of a few dial-in MUDs around 84. I think BT invested in one at some point, but that would have been later. 86, 87? Wonder how much revenue they got from that?
You have it lucky. For many, it doesn't matter whether or not company resources were used. Everything you do during the period in which you're working for them is up for grabs. At least AOL seem to make it clear where they stand.
The GPL does not allow code to be incorporated into proprietary software; theoretically the code can be relicensed to allow this each time it is needed, but to do so would require the consent of everyone who ever contributed anything to the codebase (even just a single character typo-fix to a message, or whatever)!
Alternative licenses can be written that allow open source freedom just as much, but also allow such relicensing. The license of NASM was such a license, although many object to the fact that code submitted under this license effectively becomes the property of an arbitrary group of `authors'.
I'm working on a new licensing scheme that fixes this problem and which is somewhat better phrased than the NASM license. You can see my work-in-progress on my web-site, The Democratic Software Foundation.
I appreciate comments, although don't promise to reply quickly!
First: there is no evidence that Gödel's theories don't apply to human intelligence just as much as they do to the automated reasoning processes that they were originally targeted at debunking.
Second: Gödel's theories only apply to deterministic systems. The moment you start including random numbers in the equations, you cannot get a result anymore. Your automated reasoning system can break out of the 'rut' that it was in, and up to the next class of problems. Doesn't this sound exactly like the way humans approach mathematics? Ask any mathematician who has struggled on a hard problem for years - the solution comes in a moment of inspiration - just another word for a random thought unlocking the secrets of the problem.
So, thinking machines must be nondeterministic. Fine; we already know how to make circuits behave in a nondeterministic manner.
An FTP authentication failure occurred while trying to retrieve the URL: ftp://ftp.surfnet.nl/pub/netscape/netscape6/englis h/6.0/windows/win32/
Squid sent the following FTP command:
PASS
and then received this reply
Login incorrect.
Use preview.
Must use preview.
> If you want to get really fly you could store :)
Although there's really no point, cos it'll be read out of disk cache on every hit, if you've got the right page out there!
> that file in a RAM disk.
Sorry, but I have to disagree here. Case in point - a major bug was recently discovered in the internationalisation functions of the C libraries for most major Unices. Linux and Solaris were both affected, and several commonly used setuid programs turned out to contain root shell vulnerabilities. glibc was patched to fix the problem within days, but AFAIK to date (about 4 weeks later) there is still no fix on Solaris other than disabling the setuid facility of these programs.
To anyone relying on these facilities, this is a fairly large step up in TCO.
(Disclaimer: this isn't intended as a dig at Sun. They do a great job, but it is tough to compete with free software hackers in terms of speed getting fixes out...)
As a DSVR client, I must say that their service is very good, except for a couple of problems: - software that attempts to bind sockets to IPADDR_ANY can get confused by their system and have conflict with the other virtual servers on their machine. I have this trouble with mysql, which just about everyone runs there; either you have to turn off networking or use a non-standard port. - you don't get a proper root account, and a lot of things don't work properly (eg admin can't access other users' files without relevant permissions). A service using the user mode kernel would (imho) be significantly better. But then the DSVR approach is lower cost (in terms of CPU time) and sufficient for almost all uses.
Dammit! He just let out the plot of my next blockbuster sci-fi epic adventure! Oops.
The TI99/4 was, I understand, the first 16 bit home computer available to the mass market (predating the IBM PC by about 2 years). The 4A was a fairly simple upgrade.
A point that a lot of people have missed is that the 'common sense' argument you point out is currently held by the EU. Under EU law, you can patent software, but only if it exhibits a 'novel technical effect'. My understanding of it is that this disallows any mathematically based or internal algorithmic patents (eg, the RSA patent or one that I heard about where somebody effectively had a patent on a binary search), but allows patents of new ways to make a system behave (eg BTs patent on something that approximates hyperlinks, as discussed recently) as long as nobody can show prior art.
The NASM licence is no more restrictive than the GPL is. In fact it allows certain freedoms that the GPL doesn't.
Yes, it is a little difficult to understand the parser. It could perhaps have been better structured. But more than 2 or 3 people have worked with it, and most seem to be able to grasp it.
Try actually reading the code, rather than applying automated tests. You'll find it really isn't that bad.
And some of the 'bad programming practices' you mention elsewhere actually make the code easier to read; the global variables are used so that thousands of pointers don't have to be passed all over the place. Most of them don't change throughout the operation of the program and merely indicate configuration options. Others change infrequently, or represent values that are effectively 'read-only environment values' for most parts of the code and are only changed in very rare circumstances.
In the end, what's the big deal about the GPL?
Just because its the licence that some big corps (ie redhat et al) like, does that mean we have to kowtow to their judgement? OSS developers should be free to grant whatever freedoms they like to the people using their software. A few of the freedoms we wanted to grant with NASM (and yes, I don't mind admitting that profit was among the motivations!) are incompatible with the GPL.
Allowing nasm code to be distributed under the GPL would render this impossible, because then code submitted wouldn't be cleared for inclusion under the nasm licence, but only under the GPL.
This is not an 'imperfection in the wording' as you assume; it is fully intentional and designed to preserve the rights of potential users of NASM whose uses are not covered by the GPL. And that doesn't only stretch to proprietory systems. It allows binary only distributions (placing copies of NASM on magazine covers who don't want to be bothered by restrictions on carrying source code as well, etc). It allows linking NASM code with binary only but free projects (there are some out there; not that any use NASM to my knowledge, but you never know).
mov eax, 0x0010 .exit
mov ebx, 0x5566
call myfunc
mov dword [esi + 0x0020], eax
jmp
Looks like assembler to me. What are you on about??
Indeed. GPL'd code cannot be incorporated into the NASM code without the permission of all of the authors of the GPL'd code. This is IMHO a weakness with the GPL, allowing rights over the code to be distributed among too many people. Truly free software should not make such restrictions. The NASM licence allows the NASM code to be used in situations where using GPL'd code would be impossible, because it allows a few people to make executive decisions concerning the copyright of the code.
As one of the originaly authors, we went through a stage shortly before we made our first public release of the nasm code where we wanted to make sure nasm was available to the widest range of people possible.
While the GPL is a fine licence, it suffers many shortcomings in our opinion. One of those is that it is riddled with free software political agenda. While we wanted nasm to be free, we also wanted it to be available for people who had non-free uses for it. Launching it simply under the GPL would have caused a mess; in cases where nasm's use is not allowed under the GPL (there have been several such instances), we would have to acquire the permission of everyone who ever contributed code in order to relicence outside of the scope of the GPL.
The NASM licence is designed to avoid this problem. It allows a small, executive group to take control of the rights of code submitted for inclusion in NASM, and to relicence them as they see fit.
In hindsight, there are other ways we could have gone that would have caused many problems.
But despite the contents of the e-mail that I received recently from Stallman, I am sure that the NASM licence does not break the GPL, and its inclusion within the licence means that other than where the rights of those who submit code for inclusion is restricted further, all other rights are extended beyond what they are in the GPL.
I'll post a few points elsewhere, where individual points have been made.