Re: [ql-users] SMSQ/E License
On 8 Nov 2002, at 22:21, Malcolm Cadman wrote: There seems to be some confusion around SMSQ/E's 'core' and its 'flavours'. It seems to me that Wolfgang has taken on the onerous task of maintaining the integrity of SMSQ/E so that it remains consistent and coherent, and that all users - irrespective of platform - get access to all the 'new' features. Yipee yey! Hooray! Somebody understand what I'm trying to do. Wolfgang - www.wlenerz.com
Re: [ql-users] SMSQ/E License
In article [EMAIL PROTECTED], Roy Wood [EMAIL PROTECTED] writes In message [EMAIL PROTECTED], Dave P [EMAIL PROTECTED] writes SNIP The issue I take with it is this notion that all versions of SMSQ/E must be identical. I think this is not in SMSQ/E's best interest because it discourages development. We never said they have to be identical just 'coherent'. Code that works on all platforms should be common and code that is specific need not. The thing is that all calls to platform specific code should be handled by the other platforms without a crash and the whole O/S should maintain integrity. This can only be done if the code is controlled and that is the post the Wolfgang has taken on. We all want development we also want it to work. There seems to be some confusion around SMSQ/E's 'core' and its 'flavours'. It seems to me that Wolfgang has taken on the onerous task of maintaining the integrity of SMSQ/E so that it remains consistent and coherent, and that all users - irrespective of platform - get access to all the 'new' features. So the 'core' remains robust and crash-proof on all platforms, whilst the 'flavours' can take advantage of the platform's own features. Hence SMSQ/E - with any new official version - will run on my Black Box QL + Gold Card, without any problems. Yet certain features will not be avialable because the platform simply doesn't support them. The same new official version of SMSQ/E will run on my Pentium PC with QPC2 v3, without any problems. Yet will be able to take advantage of certain new features. ... and so on for various hardware/software combinations. Perhaps this warrants a full article in QL Today, and a brief posting on this list. -- Malcolm Cadman
Re: [ql-users] SMSQ/E License
Before replying to Phoebus' post, I'd just like to say that I have the utmost respect for someone who changes their mind after expressing a view for so long. On Thu, 7 Nov 2002, [windows-1253] Öïßâïò Ñ. Íôüêïò wrote: If someone has copies of the list in-or-around '98 he will remember that I had a huge discussion with Roy where we agreed on dissagreeing on what open software is all about and why it should be supported, I did use at that time the same analogy Tony Tebby used (the author and the book) now... I like the intent of the new license - it's just a couple of specific areas I have trouble with. The license fee isn't one of them and never has been. The issue I take with it is this notion that all versions of SMSQ/E must be identical. I think this is not in SMSQ/E's best interest because it discourages development. For example, I think it is good for a version to add a feature that may not be supported by other platforms, *as long as it is an addition*, and the software style guide states that if the feature is used in software released for all platforms, the equivalent functionality should be included (if possible) for other platforms too. For example, say a machine is released that requires different code to operate an IDE interface. That version should be allowed to exclude code which is simply not relevant, like microdrive-related code, if microdrives could never be attached to the machine. (this may be a bad example) That IDE code may for hardware reasons be entirely irrelevent to every other version, but be required for this version just to achive the same functionality. Dealing with the machine at a hardware level, it seems silly to require that all those patches be included in all versions of SMSQ/E, and/or that additional features to support extra hardware be globally applied even if other machines could never support the hardware. I hope I explained this properly - it's a difficult thing to explain when I'm tired and can't find the words. Dave
Re: [ql-users] SMSQ/E License
On 07/11/02 at 17:07 Dave P wrote: The issue I take with it is this notion that all versions of SMSQ/E must be identical. I think this is not in SMSQ/E's best interest because it discourages development. For example, I think it is good for a version to add a feature that may not be supported by other platforms, *as long as it is an addition*, and the software style guide states that if the feature is used in software released for all platforms, the equivalent functionality should be included (if possible) for other platforms too. For example, say a machine is released that requires different code to operate an IDE interface. That version should be allowed to exclude code which is simply not relevant, like microdrive-related code, if microdrives could never be attached to the machine. (this may be a bad example) Then what you really want to say in the licence would be that additions to SMSQ/E to add a feature of capability to one platform will not be accepted if it may seriously hinder or even prevent adding an equivalent feature or capability on another. Obviously, this excludes all platforms where such feature simply makes on sense or is impossible (by design - leack of need), but does at least suggest some form of forethought, so that we don't get 'my way or the highway' style features. This breeds tremenodous problems with writing applications and further additions to SMSQ/E. Nasta
Re: [ql-users] SMSQ/E License
On Thu, 7 Nov 2002, ZN wrote: Obviously, this excludes all platforms where such feature simply makes on sense or is impossible (by design - leack of need), but does at least suggest some form of forethought, so that we don't get 'my way or the highway' style features. This breeds tremenodous problems with writing applications and further additions to SMSQ/E. Or does it? My vision is: Someone writes a spiffy new [item] and it gets included on that version of SMSQ, because [XYZ] couldn't/wouldn't support it anyway. The feature is now available if people need it, and they can write applications (remember when we used to call them programs?) that can benefit from the feature. Now, your average software author writing for the market can use the feature if he wants to, but wouldn't want to limit his market to that platform alone. This is an *operating system* we're talking about, a way for software to use the hardware. If code running on radically different hardware cannot be modified to take into account features of that specific hardware, that really limits the development of hardware, and of software that runs on it. The notion that all versions should have identical features could be replaced with the notion that all versions should have compatible features, even if the capability is different. EG: an ARM-QL could support 1600x1200 on a CRT, or 2048x1640 on a LCD. With touch screen support. And USB. And ethernet. Things that should be included in a monolithic OS. imvvvho Dave
Re: [ql-users] SMSQ/E License, Metadrivers and other nice stories
??? 7/11/2002 12:07:12 ??, ?/? Dave P [EMAIL PROTECTED] ??: Before replying to Phoebus' post, I'd just like to say that I have the utmost respect for someone who changes their mind after expressing a view for so long. Thanks. Please here note that if TT had made a different point in his email my objections would still stand on the license... It's just that up until now, I regarded the SMSQ/E license under a different light. Truth be told that a program developed continuously since 1983 at a certain point stops being a product and becomes a labour of love (it's art in its own right). This fact I overlooked. Regarding the rest of the OSS world however, I stand by my original opinion, which isn't that much unlike Dave's :-) snip The issue I take with it is this notion that all versions of SMSQ/E must be identical. I think this is not in SMSQ/E's best interest because it discourages development. I would disagree on that. I don't believe that the license's intent to make everything identical. It's intent as I perceive it to make non-hardware specific functions behave the same. In that sense, if you are writing a program that's graphics oriented (to come back to my favourite subject), its useless if on a specific platform the same calls do different things. In the case of graphics for example I would say that you are confusing functionality with output. SMSQ/E achieves that as it stands now in the sense that even non-colour aware versions of SMSQ/E (later ones granted) can understand the colour specific commands even if the hardware (or the specific version's drivers) can't display them. That's not bad. It does ensure a level of compatibility which is not perfect but better than nothing Before something jumps in and says that I said to Roy that if totally new features were introduced that made older software incompatible that was okay... please note that these features should be available EVERYWHERE... therefore making SMSQ/E for ALL platforms incompatible, not just one :-) For example, I think it is good for a version to add a feature that may not be supported by other platforms, *as long as it is an addition*, and the software style guide states that if the feature is used in software released for all platforms, the equivalent functionality should be included (if possible) for other platforms too. Exactly my point :-) And also at least the other be aware of the extra functionality and not crash... This cannot be possible with patches (which in the specific Q60 case may or may not be affecting code functionality on other platforms, but that's besides the point here) but only with incorporation in the main source tree... For example, say a machine is released that requires different code to operate an IDE interface. That version should be allowed to exclude code which is simply not relevant, like microdrive-related code, if microdrives could never be attached to the machine. (this may be a bad example) That IDE code may for hardware reasons be entirely irrelevent to every other version, but be required for this version just to achive the same functionality. See here you are partially wrong and I will tell you why too. (I owe this explanation to Nasta btw) The problem with QDOS/SMSQ so far (and I believe its the most serious problem) is that in reality its drivers are really custom pieces of code (as in code that someone would write only to support a specific hardware/software combination and that wouldn't be available elsewhere in the OS) that really have nothing to do with the OS itself (the famous metadrivers discussion again). While QDOSMSQ provides us with the necessary functionality to do these things (and more... Really if we think about something new chances are that SMSQ/E already has it... problem is Tony never really told us about it... but more on that in a second), no one really sat down to write a driver framework (Objects, Methods and Classes... yes in reality SMSQ/E is a truly Object Oriented OS) and thus eliminating the need to keep drivers in the source tree. If such a framework is introduced, first of all, all discussion on how drivers and hardware specific functionality is implemented will immediately cease. (A very basic level of a BIOS equivalent with a quasi-driver of course will have to be maintained in the source tree in order for one to be able to boot the machine... that's understandable but the drivers in that case being what they should be wouldn't cause either controversy or incompatibilities ;-). Dealing with the machine at a hardware level, it seems silly to require that all those patches be included in all versions of SMSQ/E, and/or that additional features to support extra hardware be globally applied even if other machines could never support the hardware. True (see my take on the subject above :-) but in any case that wouldn't matter... as the interface would be there, just not the drivers... so you call
Re: [ql-users] SMSQ/E License
??? 7/11/2002 1:02:47 ??, ?/? ZN [EMAIL PROTECTED] ??: snip Then what you really want to say in the licence would be that additions to SMSQ/E to add a feature of capability to one platform will not be accepted if it may seriously hinder or even prevent adding an equivalent feature or capability on another. Obviously, this excludes all platforms where such feature simply makes on sense or is impossible (by design - leack of need), but does at least suggest some form of forethought, so that we don't get 'my way or the highway' style features. This breeds tremenodous problems with writing applications and further additions to SMSQ/E. Nasta While your original proposition (ca. 1995 IQLR) would already have solved the problem... Not to mention that with the exception of writing a Memory Manager, a Driver API or an internal IP stack (for other purposes eg. making it an embedded OS for Internet devices and even that wouldn't really qualify), nothing extremely new can be done to improve SMSQ/E which is extremely robust as it is... Mind you that I believe that the following are not (and should not) be part of the OS core: Drivers (keyboard/screen/serial/disk/sound etc.) S*Basic PE Command line I/F (although for obvious -and historical- reasons these are included and perceived as a part of the OS and this illusion isn't really bad) File System (although some may disagree, but for me the FS isn't nothing more than the root object on where other objects (devices) are attached - Again SMSQ/E does provide object functionality, just doesn't call it that ;-)) IMHO what SHOULD BE part of the os ONLY: Scheduler I/O API (The drivers framework) Housekeeper (process manager) Job communication (Call them pipes or what ever you like) Object Manager (The Thing system) Phoebus
Re: [ql-users] SMSQ/E License
In message [EMAIL PROTECTED], Dave P [EMAIL PROTECTED] writes SNIP The issue I take with it is this notion that all versions of SMSQ/E must be identical. I think this is not in SMSQ/E's best interest because it discourages development. We never said they have to be identical just 'coherent'. Code that works on all platforms should be common and code that is specific need not. The thing is that all calls to platform specific code should be handled by the other platforms without a crash and the whole O/S should maintain integrity. This can only be done if the code is controlled and that is the post the Wolfgang has taken on. We all want development we also want it to work. -- Roy Wood Q Branch, 20 Locks Hill Portslade. Sussex. BN41 2LB. UK Tel : +44 (0)1273 386030 Fax : +44 (0)1273 430501 (New number!) Mobile +44(0)7836 745501 Web : www.qbranch.demon.co.uk
Re: [ql-users] SMSQ/E License
On 7 Nov 2002, at 17:07, Dave P wrote: (different possibilities on different machines) I don't know what I'm doing wrong, but, obviously it is something. I distinctly seem to remember that, as long as something is useful only for one machine, I see no problem in putting it in the code for that machine. COnversely, it does without saying that machines that don't have a facility (your example of mdvs is great) then they don't have it - Q60 or the Atari don't have mdv code and why should they? On the other hand, if some new functionality exists for a machine which can be useful for other platforms, why exclude it? Wolfgang
Re: [ql-users] SMSQ/E License
John Sadler wrote: All your problems would be solved if you use the LGPL license, if the soure and code is going to be free. Anybody would be able to sell comercial programs using the updated SMSQ/E code. Official versions would still have to be ratified by the appropiate person. Yes, I think LGPL would do, but even that isn't necessary. Under GPL, the copyright holder defines what *linking* means. E.g. he could add this to the GPL copyright notice: As a special exception, if *other* files instantiate templates or use macros or inline functions from this file, or you compile this file and link it with other works to produce a work based on this file, this file does not by itself cause the resulting work to be covered by the GNU General Public License. However the source code for *this* file must still be made available in accordance with section (3) of the GNU General Public License. This way an emulator software, or other parts of SMSQ/E could be *strictly* commercial, could use all the templates they need, without falling under the GPL. Moreover, it could even be linked into the same binary or ROM. (Which is not necessary in our case.) That's how it's done in embedded computing. Peter
