I had a major breakdown (someone fiddled around with the server, filling the disk), but it's up again (damn, why does it break just today:}). Unfortunately some newer comments were lost. I hope I can get them back, and then I'll move to Zope 2;)
The login prompt was due to wrong TinyTable permissions after the backup.
Not listed on Mattshouse, but nearly what the guy is looking for is KMySQL. Despite its name, the recent versions of KMySQL have a plugin architecture with support for PostgreSQL and mysql. It would be worthwhile to add ORACLE support by writing a plugin
Re:CORBA can be fast - and incompatible...
on
KDE Looks Ahead
·
· Score: 1
I don't want to nitpick, but GNOME's CORBA implementation is not exactly kosher.
Usually using a different ORB isn't a problem, even if it has some custom features which make switching ORBs a bit more difficult. The Internet Inter ORB Protocol (IIOP) ensures that ORBs can speak to each other.
Now GNOME uses an authentication mechanism that simply breaks the IIOP, and makes interoperability with other ORBs impossible (ORBit does by far not support the whole CORBA standard).
This defeats the purpose of having CORBA at all.
Enlighten me if this has changed already. Right now the KDE way is much more sensible: - use CORBA properly, where it makes sense - use shared libs where speed is required
It would also be possible to have database forms and views that adapt to the current desktop, using the XML tags stored in the database. At least for simpler, not performance-critical GUIs this is an option.
Just, as the above poster said, find someone to write it;-)
Even in $, it would be reasonable. Look at the insane prices for other IT comoanies, not to mention events like the RedHat IPO, where a brand name (there isn't much more considering the turnover) brought $5bil. the reasons:
StarOffice practically own the German private market, and this is very attractive, as German companies are very unsatisfied with MS and more likely to switch than their US couterparts. And then, it's the largest market after the US.
StarOffice rides the Linux wave, and pretty big. They had a head start to Corel, and they've greatly benefited from that. It's fair to day StarOffice is the office suite for Linux currently
They're the only ones who have managed to build a sensible network-based productivity application using Java. Sure, IBM has a lot of specialized stuff, but Lotus' Java suite is pathetic and Corel has given up. So if Sun wants Java to succeed in that market, they just had to look at StarDivision.
SO is available for Solaris. Maybe the only usable office tool?
StarDivision is probably a very healthy company. SUN would have liked to buy it much earlier, but out of a contract with IBM just before the Notes aquisition they got a lot of money from IBM breaking the contract (~15m is assumed). That's why they could market their product so aggressively.
All in all maybe not a bargain, but not that much either, considering that SUN had little choice but to buy them. They would have relied on StarOffice anyway.
This system is IMHO overkill, considering the incident that triggered it (3 spam accounts). It may also be in part unjust and harmful to the quality. The Karma system is a cool idea, but it is too general at the moment. It will create 'hero worshipping' in that frequent posters and known Linux personalities will immediately have a high score, even if they just wanted to make a light-hearted remark or are not so qualified at the particular topic. This could take the spontaneousness away. My Suggestions: A logged-in user should *never* start lower than an AC If s/he slips below 0, a mail should be sent to him explaining his behaviour. With 3-5 rogue accounts, it shouldn't be difficult to sort them out manually, and spammers like 'i hate stevens' won't bother to justify themselves.
Karma should be constantly updated: as others suggested, based on a time span (last 2 months) and/or number of posts
Karma should be restricted to topic. This would be tough to implement, but would reflect that the respective poster is qualified at a cetain (range of) topic(s).
The main problem are IMHO not the registered users, but systematic abuse of AC. I really enjoy AC opinions sometimes, but recently it seems they have been abused to stir up a lot of trouble. Maybe MS anti-Linux Division,/.-trolls department??
Well, you can have it either way. As long as you want to make a weblog or bulletin board news site, you will always end up with something in between Linuxtoday, Linux.com, Slashdot, or the big TV sites (CNN,BBC).
The KDE Community Forum for example looks roughly like the other KDE pages ('Community identity'), but keeps the/.-like header icons (coll thing for an overview).
The importance is to fill it with life and good information, the packaging is just one part.
Join the KCF - where the K community maps to Z objects!
I may be biased, but....
if you're into software development full time, USD 1000 will pay for your time and expenses for several DAYS.
Unfortunately, you are not only biased, but also wrong. It would still be nice if this statement were true though:)
So what's exactly wrong with this? If you don't engage in slave labour, you'll get around $4k a month, so $1k would be a week of work. When you take into account that you get 1 year of support for this money, this is very little.
Or take the shareware case: I presume you are still thinking in the dimensions of the Windows world, where cheap shareware is very common. On UNIX, with all the powerful free tools, you won't be able to sell anything except rather sophisticated tools, say for ca. $100. So you'd have to sell 10 pieces for the license cost.
(Note: TrollTech doesn't offer a shareware license for administration reasons. I guess a Qt version bundled with Delphi would be much cheaper).
There is nothing in OO concepts that can't be expressed in C or any other programming language.
You can't enforce data hiding and encapsulation, many kinds of compile-time type checking (let alone RTTI), no type-save, generic templates or containers, just macros that are processed by the preprocessor, thus making automatic type checking impossible and debugging very unpleasant (references are to the precompiler output, not the actual source code).
Moreover, abstractions like overloaded functions and operators aren't possible.
OO is a mindset, not a language feature.
Sure, but unless you are ingenious, know all your structures and object hierarchies by heart and are not coding in a team, it is vital that OO be supported by language features.
It's just like you can do anything by writing zeros and ones as well...
It's once again the topic of "Why should development tools be free when the OS is proprietary?" It's a nuisance for a small fraction of developers (those who are interested in developing cross-platform open-source GUI apps), but most people just don't care. And then,you can still implement the GUI part (usually lt 10% code) in, say, MFC, like AbiWord.
No, they realy want the money, and won't let you use Windows Qt without paying $1000+. I'd guess writing native Win32 support is quite an additional effort, and thus probably worth the money. One possible short-term solution: Making Qt apps work on a free Windows X server. This shouldn't be too much effort, and AFAIK the QPL doesn't fobid that either. Does anyone know of a free X server coming with devel libs (MI/X is AFAIk binary only, and XFree not available for Windows)?
Those standards are changing pretty fast at the moment. With Open Source Unix variants constantly gaining market share, the importance of CDE+Motif decreases. Qt seems to be used by rather large companies (see this interview), so there is obviously room besides Motif.
In the end, companies care about productivity in programming, and that's an area where Qt definitely beats Motif. The only question is how much the underlying toolkit (e.g. Motif vs Qt) would affect coding in Delphi/Pascal.
That's indeed an interesting point: Theory and Real Life seem to be further apart in the US than elsewhere.
You all love "Life, Liberty and the Pursuit of H." (soken with a Reaganesque accent), but how does reality look like?
Insane Legal system combined with love for death penalty mean that you can be sure of your life unless you have enough money to afford a real good lawyer. Faking evidence to make you look guilty is always possible, as long as enough money is involved.
Liberty is cool, as long as you do what the majority does and what's Politically Correct. If you dress/behave differently as a teenager (->thrown out of school), or drink beer at your marriage when below 21 (->thrown into jail), have a non-capitalist/nationalist attitude (->McCarthy-style persecution->no job), then bad luck to you... At least you can drive buy a machine gun at 16...
Happiness is a great goal, but don't you ask for help when your not happy, say your lungs collapsed (->no health insurance), your neigbor doesn't like you (-> can't affort lawyer) or you've served coffee too hot (->$2 million fine), or, heavens, lose your job or get a child without being happily married. Don't you ask for help then! Life and happiness are your own business.
The main difference to China is that a) Chinese convicts have to pay for their capital punishment themselves, b) the communist party conveniantly defines more clearly what 'politically correct' means and c) disputes are handled by paying bribes directly instead to the lawyers...
Unlikely to happen. Not only because of MS anti-Linux stance, but also because COM/OLE components would be not a pretty hack on Linux (not that's impossible, Software AG already has something similar).
Try QtEZ or KDevelop, if you can handle C++, otherwise look forward to KOM/OpenParts components, esp. with Python support.
GUI toolkit is the important question!
on
Delphi for Linux
·
· Score: 1
Delphi itself is just another fine IDE. By the time Delphi for Linux will be released, we'll have quite a number of IDEs anyway.
The crucial thing is what widget set Borland/Inprise will use.
They could write one of their own (partly using Win code) or use wrapper around existing toolkits. The conservative (and IMHO stupid) solution would be to take Motif. The cheapest solution would be a free toolkit like fltk or, more likely GTK+, although Inprise could be hesitant to embrace a fast moving open source 'target'. A very interesting solution would be Qt. The main complaint at the moment is the price, which makes it more difficult to develop small commercial or shareware programmes on Linux/Unix. This is due to the fact that you get 1 year of support from TrollTech.
A cheaper, no-support version bundled with Delphi would definitely be attractive for many developers, while the high number of sales would ensure TrollTech the necessary revenue.
c't is the authoritative source for geeks and nerds (who understand German or Dutch) and way more independent and credible than that ZiffDavis crap like PCWeek.
From the c't article "As is the custom with Microsoft, the introduction of this new proprietary standard..." sounds like they didn't do their homework since this is an open standard, Nullsoft didn't seem to have any problem adding support for ASF to Winamp. Garbage. Nullsoft has to license it like everybody else. Where's the source, where's the (OSS) encoder? ASF is absolutely proprietary. MP3, while having some patent woes, can be encoded with an open source product like LAME.
PCWeek says that ASF sounds as good or better than MP3 There is no backup of that claim whatsoever, it's just a short note, so it's highly unlikely that they tested it properly (it wasn't even released by then). The c't article OTOH presents an excellent test complete with audio samples and precise descriptions where ASF fails miserably in comparison to MP3 whith distortions and a lower signal resolution. It is, however, comparable to other pure streaming formats like Real Audio.
The number of trolling MS employees on/. is getting annoying recently, I must say...
You don't have to use Yast if you don't like it. Just install Linuxconf or hack the files yourself. The best thing about SuSE is the quality and sheer number of the included packages, as well as the SaX XFree setup tool (GPL'd, in case you're interested).
And YaST's license is no big deal either. Source is available, modifiable, distributable. You just have to ask if you want to sell it for more than the media cost (so the 2$ CD's are usually OK). Thats nothing else than with mySQL or the Cyrus IMAP server.
While it would be best to have some X experts like Raster or Mosfet comment on the subject, I think a few things are quite obvious to the end user.
X has similar backwards compatibility problems as Windows: There is support for many things like strange old font formats that are not necessary anymore. Often the code itself is quite old (pre-ANSI) and hard to read.
Lack of modularity: At least XFree is just a big junk of code that doesn't allow run-time modules to be loaded. Result: Bloat and little flexibility. Apparentlt XFree 4 will change this.
Lacks essential features like Anti-Aliasing, TrueType support requires additional software
Optimized for networking, not 'local' desktops: Performance critical apps like games or video need hacks that are either dangerous (suid root) or not available everywhere
For networking: Rather inefficient protocol, doesn't support compression (without proxie tools) or encryption of a session
'Understandardized': No modern GUI toolkit included -> many different looks&feels, Motif as perceived 'de-facto' standard expensive and weak.
No component model: There is still no way to use e.g. CORBA and X-based components over a network, as there is no authentication mechanism available.
Implementation language and coding style: A lot of the code was written when there was no ANSI-C, let alone C++. So code is somewhat difficult to read and the object oriented nature of X is achieved only with a complex and fragile (=error-prone) construction in C.
Risk for stability: As X has direct acess to the hardware, it is as stability-critical as the kernel, for desktop users. Moreover, the kernel has no control over X memory management, and X doesn't 'shrink', so it's actually a kind of memory leak...
Alternatives There is the GGI-Project, the Berlin consortium and Y. GGI is an abstraction layer for grafics programming, but relatively low-level (e.g.games). It serves as driver lib, can be partially controlled by the kernel (KGI) and can output to different "targets" like svga or X. It also serves as the basis for Berlin, a new windowing system using OpenGL, among other things. Don't know about Y.
He was probably referring to this mail by Holger Kruse, author of the Miami TCP/IP stack and GUI for Amiga. I don't know if his opinion concerning Linux problems with newer Internet features is right, but it sounds relatively credible (at least it did to LeFaivre).
Your informatio about the Qt license is obsolete. The patchwork text isn't applicable to the QPL 1.0. CVS development is possible, and so is forking and changes in the code. The requirement to mark those changes clearly is the same as in the GPL.
Actually the only "problem" with the QPL in Qt's case is not one of free software: It's the fact that shareware authors have a tougher life. That's why they are choosing LGPL'd libraries like fltk or gtk. In Zope's case it remains to be seen if there is a commercial license for proprietary modules, and how much it will be.
Abd BTW, if you like python, there is Zope as well.
One of the reasons TrollTech stated for not taking the GPL (+ an additional commercial license) was this legal aspect. RedHat stated this assumption as one of the main risks when they filed for the IPO. Many other companies and press reviewers mention that point. So this seems to be a valid concern. But why?
The GPL is both long *and* unclear. It tries to be comprehensive and precise but fails miserably. Think of the discussions about the meaning of (the current version of) the GPL. Everyone disagrees with each other, and in the end the dispute is settled by asking RMS. So the GPL essentially means what RMS wants it to mean. This may be OK in a moral context, but it is insufficient for legal purposes. The main problems with the GPL v2: -> No clear vision of "linking" and "derived works": Todays software words sees much more kinds of dynamic cooperation between software objects than a couple of years ago. Plugin architectures, CORBA components, XML based GUI apps etc etc, make the distinction between derived/combined and independent works VERY tricky. -> The "system component" clause is extremely unclear. Who defines what's part of the system? What ISN'T part of the system, when distributors ship gigabytes of software with their "system"? This has to be clarified IN the license, not by the whim of its creator. -> "GPL compatibility": In short, a license is GPL compatible when RMS says so. Period. There are rules of thumb (more/less restrictive), but none of these are justified by the license text. If you like, I can give you very valid reasons why the X license is not GPL compatible...
In short, lawyers, at least here in continental Europe, laught at you if you show them that peculiar license text (that's what I have been told). Caldera's and TrollTech's lawyers obviously felt the same.
In fact, the QPL is a pretty good license, if you like the GPL. It's essentially GPL plus right to link to all other Open source code minus right to do 'closed' in-house development.
AFAI understand it, those restrictions do only apply to modifications of the Zend engine, not to scripts in general. As far as you don't link with the Zope code, this shouldn't be a problem, if you apply the analogy to Qt. 'Linking' would probabably mean 'building compiled modules', which you could not distribute without source.
One reason for choosing the QPL could be that the GPL is a) not compatible with (many) other Open Source licences and b) unlikely to be defensible in court.
But, if you need a python-based web publishing environment, you could as well take Zope
Slashdot Mirror
I had a major breakdown (someone fiddled around with the server, filling the disk), but it's up again (damn, why does it break just today :}). ;)
Unfortunately some newer comments were lost. I hope I can get them back, and then I'll move to Zope 2
The login prompt was due to wrong TinyTable permissions after the backup.
Not listed on Mattshouse, but nearly what the guy is looking for is KMySQL.
Despite its name, the recent versions of KMySQL have a plugin architecture with support for PostgreSQL and mysql.
It would be worthwhile to add ORACLE support by writing a plugin
I don't want to nitpick, but GNOME's CORBA implementation is not exactly kosher.
Usually using a different ORB isn't a problem, even if it has some custom features which make switching ORBs a bit more difficult. The Internet Inter ORB Protocol (IIOP) ensures that ORBs can speak to each other.
Now GNOME uses an authentication mechanism that simply breaks the IIOP, and makes interoperability with other ORBs impossible (ORBit does by far not support the whole CORBA standard).
This defeats the purpose of having CORBA at all.
Enlighten me if this has changed already. Right now the KDE way is much more sensible:
- use CORBA properly, where it makes sense
- use shared libs where speed is required
I do second that.
;-)
It would also be possible to have database forms and views that adapt to the current desktop, using the XML tags stored in the database.
At least for simpler, not performance-critical GUIs this is an option.
Just, as the above poster said, find someone to write it
---
Unlikely. The maximum allowed is 2.5 kg (vs 16 kg).
...
Even litres vs gallons wouldn't suffice.
I wonder if this was software controlled
the reasons:
It's fair to day StarOffice is the office suite for Linux currently
All in all maybe not a bargain, but not that much either, considering that SUN had little choice but to buy them. They would have relied on StarOffice anyway.
This system is IMHO overkill, considering the incident that triggered it (3 spam accounts).
/.-trolls department??
It may also be in part unjust and harmful to the quality.
The Karma system is a cool idea, but it is too general at the moment. It will create 'hero worshipping' in that frequent posters and known Linux personalities will immediately have a high score, even if they just wanted to make a light-hearted remark or are not so qualified at the particular topic. This could take the spontaneousness away.
My Suggestions:
A logged-in user should *never* start lower than an AC
If s/he slips below 0, a mail should be sent to him explaining his behaviour. With 3-5 rogue accounts, it shouldn't be difficult to sort them out manually, and spammers like 'i hate stevens' won't bother to justify themselves.
Karma should be constantly updated: as others suggested, based on a time span (last 2 months) and/or number of posts
Karma should be restricted to topic. This would be tough to implement, but would reflect that the respective poster is qualified at a cetain (range of) topic(s).
The main problem are IMHO not the registered users, but systematic abuse of AC. I really enjoy AC opinions sometimes, but recently it seems they have been abused to stir up a lot of trouble.
Maybe MS anti-Linux Division,
Well, you can have it either way. As long as you want to make a weblog or bulletin board news site, you will always end up with something in between Linuxtoday, Linux.com, Slashdot, or the big TV sites (CNN,BBC).
/.-like header icons (coll thing for an overview).
The KDE Community Forum for example looks roughly like the other KDE pages ('Community identity'), but keeps the
The importance is to fill it with life and good information, the packaging is just one part.
Join the KCF - where the K community maps to Z objects!
So what's exactly wrong with this? If you don't engage in slave labour, you'll get around $4k a month, so $1k would be a week of work. When you take into account that you get 1 year of support for this money, this is very little.
Or take the shareware case:
I presume you are still thinking in the dimensions of the Windows world, where cheap shareware is very common.
On UNIX, with all the powerful free tools, you won't be able to sell anything except rather sophisticated tools, say for ca. $100. So you'd have to sell 10 pieces for the license cost.
(Note: TrollTech doesn't offer a shareware license for administration reasons. I guess a Qt version bundled with Delphi would be much cheaper).
There is nothing in OO concepts that can't be expressed in C or any other programming language.
You can't enforce data hiding and encapsulation, many kinds of compile-time type checking (let alone RTTI), no type-save, generic templates or containers, just macros that are processed by the preprocessor, thus making automatic type checking impossible and debugging very unpleasant (references are to the precompiler output, not the actual source code).
Moreover, abstractions like overloaded functions and operators aren't possible.
OO is a mindset, not a language feature.
Sure, but unless you are ingenious, know all your structures and object hierarchies by heart and are not coding in a team, it is vital that OO be supported by language features.
It's just like you can do anything by writing zeros and ones as well...
It's once again the topic of "Why should development tools be free when the OS is proprietary?"
It's a nuisance for a small fraction of developers (those who are interested in developing cross-platform open-source GUI apps), but most people just don't care. And then,you can still implement the GUI part (usually lt 10% code) in, say, MFC, like AbiWord.
No, they realy want the money, and won't let you use Windows Qt without paying $1000+.
I'd guess writing native Win32 support is quite an additional effort, and thus probably worth the money.
One possible short-term solution: Making Qt apps work on a free Windows X server. This shouldn't be too much effort, and AFAIK the QPL doesn't fobid that either.
Does anyone know of a free X server coming with devel libs (MI/X is AFAIk binary only, and XFree not available for Windows)?
Those standards are changing pretty fast at the moment.
With Open Source Unix variants constantly gaining market share, the importance of CDE+Motif decreases.
Qt seems to be used by rather large companies (see this interview), so there is obviously room besides Motif.
In the end, companies care about productivity in programming, and that's an area where Qt definitely beats Motif.
The only question is how much the underlying toolkit (e.g. Motif vs Qt) would affect coding in Delphi/Pascal.
That's indeed an interesting point: Theory and Real Life seem to be further apart in the US than elsewhere.
You all love "Life, Liberty and the Pursuit of H." (soken with a Reaganesque accent), but how does reality look like?
Insane Legal system combined with love for death penalty mean that you can be sure of your life unless you have enough money to afford a real good lawyer. Faking evidence to make you look guilty is always possible, as long as enough money is involved.
Liberty is cool, as long as you do what the majority does and what's Politically Correct.
If you dress/behave differently as a teenager (->thrown out of school), or drink beer at your marriage when below 21 (->thrown into jail), have a non-capitalist/nationalist attitude (->McCarthy-style persecution->no job), then bad luck to you...
At least you can drive buy a machine gun at 16...
Happiness is a great goal, but don't you ask for help when your not happy, say your lungs collapsed (->no health insurance), your neigbor doesn't like you (-> can't affort lawyer) or you've served coffee too hot (->$2 million fine), or, heavens, lose your job or get a child without being happily married.
Don't you ask for help then! Life and happiness are your own business.
The main difference to China is that
a) Chinese convicts have to pay for their capital punishment themselves,
b) the communist party conveniantly defines more clearly what 'politically correct' means and
c) disputes are handled by paying bribes directly instead to the lawyers...
OK, enough of that rant.
Unlikely to happen. Not only because of MS anti-Linux stance, but also because COM/OLE components would be not a pretty hack on Linux (not that's impossible, Software AG already has something similar).
Try QtEZ or KDevelop, if you can handle C++, otherwise look forward to KOM/OpenParts components, esp. with Python support.
Delphi itself is just another fine IDE. By the time Delphi for Linux will be released, we'll have quite a number of IDEs anyway.
The crucial thing is what widget set Borland/Inprise will use.
They could write one of their own (partly using Win code) or use wrapper around existing toolkits.
The conservative (and IMHO stupid) solution would be to take Motif.
The cheapest solution would be a free toolkit like fltk or, more likely GTK+, although Inprise could be hesitant to embrace a fast moving open source 'target'.
A very interesting solution would be Qt. The main complaint at the moment is the price, which makes it more difficult to develop small commercial or shareware programmes on Linux/Unix. This is due to the fact that you get 1 year of support from TrollTech.
A cheaper, no-support version bundled with Delphi would definitely be attractive for many developers, while the high number of sales would ensure TrollTech the necessary revenue.
Sorry you WinTroll, you're talking garbage!
/. is getting annoying recently, I must say...
c't is the authoritative source for geeks and nerds (who understand German or Dutch) and way more independent and credible than that ZiffDavis crap like PCWeek.
From the c't article "As is the custom with Microsoft, the introduction of this new proprietary standard..." sounds like they didn't do their homework since this is an open standard, Nullsoft didn't seem to have any problem adding support for ASF to Winamp.
Garbage. Nullsoft has to license it like everybody else. Where's the source, where's the (OSS) encoder? ASF is absolutely proprietary.
MP3, while having some patent woes, can be encoded with an open source product like LAME.
PCWeek says that ASF sounds as good or better than MP3
There is no backup of that claim whatsoever, it's just a short note, so it's highly unlikely that they tested it properly (it wasn't even released by then).
The c't article OTOH presents an excellent test complete with audio samples and precise descriptions where ASF fails miserably in comparison to MP3 whith distortions and a lower signal resolution. It is, however, comparable to other pure streaming formats like Real Audio.
The number of trolling MS employees on
You don't have to use Yast if you don't like it. Just install Linuxconf or hack the files yourself.
The best thing about SuSE is the quality and sheer number of the included packages, as well as the SaX XFree setup tool (GPL'd, in case you're interested).
And YaST's license is no big deal either. Source is available, modifiable, distributable.
You just have to ask if you want to sell it for more than the media cost (so the 2$ CD's are usually OK). Thats nothing else than with mySQL or the Cyrus IMAP server.
It's here.
;^)
How could you dare to think there is something that's not on the SUSE CDs
BTW, SuSE has hired the main ALSA developer and hosts the project homepage.
So I guess they will be the driving force behind ALSA in the future...
The appropriate c't article via Babelfish (German->English) is here:
Athlon article
Unfortunately Babelfish translates only half of it...
Alternatives
There is the GGI-Project, the Berlin consortium and Y. GGI is an abstraction layer for grafics programming, but relatively low-level (e.g.games). It serves as driver lib, can be partially controlled by the kernel (KGI) and can output to different "targets" like svga or X. It also serves as the basis for Berlin, a new windowing system using OpenGL, among other things. Don't know about Y.
Hope that helps a bit
Avus
KDE uses the 'single-click paradigm' pretty consistently, which is IMHO a good thing. Once you get used to it, you won't want to miss it.
The double-click folks should try out KExplorer (now called Kruiser), which is a nice MS Explorer clone which behaves in the 'traditional' way.
He was probably referring to this mail by Holger Kruse, author of the Miami TCP/IP stack and GUI for Amiga.
I don't know if his opinion concerning Linux problems with newer Internet features is right, but it sounds relatively credible (at least it did to LeFaivre).
Your informatio about the Qt license is obsolete. The patchwork text isn't applicable to the QPL 1.0.
CVS development is possible, and so is forking and changes in the code. The requirement to mark those changes clearly is the same as in the GPL.
Actually the only "problem" with the QPL in Qt's case is not one of free software: It's the fact that shareware authors have a tougher life. That's why they are choosing LGPL'd libraries like fltk or gtk.
In Zope's case it remains to be seen if there is a commercial license for proprietary modules, and how much it will be.
Abd BTW, if you like python, there is Zope as well.
Could you please elaborate more on that?
:)
OK. You asked for it
One of the reasons TrollTech stated for not taking the GPL (+ an additional commercial license) was this legal aspect.
RedHat stated this assumption as one of the main risks when they filed for the IPO.
Many other companies and press reviewers mention that point.
So this seems to be a valid concern. But why?
The GPL is both long *and* unclear. It tries to be comprehensive and precise but fails miserably.
Think of the discussions about the meaning of (the current version of) the GPL. Everyone disagrees with each other, and in the end the dispute is settled by asking RMS. So the GPL essentially means what RMS wants it to mean. This may be OK in a moral context, but it is insufficient for legal purposes.
The main problems with the GPL v2:
-> No clear vision of "linking" and "derived works": Todays software words sees much more kinds of dynamic cooperation between software objects than a couple of years ago. Plugin architectures, CORBA components, XML based GUI apps etc etc, make the distinction between derived/combined and independent works VERY tricky.
-> The "system component" clause is extremely unclear. Who defines what's part of the system? What ISN'T part of the system, when distributors ship gigabytes of software with their "system"? This has to be clarified IN the license, not by the whim of its creator.
-> "GPL compatibility": In short, a license is GPL compatible when RMS says so. Period. There are rules of thumb (more/less restrictive), but none of these are justified by the license text. If you like, I can give you very valid reasons why the X license is not GPL compatible...
In short, lawyers, at least here in continental Europe, laught at you if you show them that peculiar license text (that's what I have been told). Caldera's and TrollTech's lawyers obviously felt the same.
In fact, the QPL is a pretty good license, if you like the GPL. It's essentially GPL plus right to link to all other Open source code minus right to do 'closed' in-house development.
AFAI understand it, those restrictions do only apply to modifications of the Zend engine, not to scripts in general. As far as you don't link with the Zope code, this shouldn't be a problem, if you apply the analogy to Qt. 'Linking' would probabably mean 'building compiled modules', which you could not distribute without source.
One reason for choosing the QPL could be that the GPL is
a) not compatible with (many) other Open Source licences and
b) unlikely to be defensible in court.
But, if you need a python-based web publishing environment, you could as well take Zope