Yes, it is possible, while you usually do not want it. It is not easy to get a server system scalable for a whole school, with service contracts, warranty and stability for this application. The central application server requires hardware that fits pretty much into what Sun offers with their machines and solutions. If you are in a university, going with Linux here would be cool, as you would have lots of students to hire for working on this. But in a school, you do not want to have more trouble then saying "please get us 25% more RAM into our server and the latest software update" to the school secretary, so that she calls and gets the service installing this in less then a work-day...
The size of your Photoshop file should not matter, as it stays on the server. And you will have a server with a lot of memory. By the way, your use case is a great example - have 100 clients, and 40 of them running Photoshop, the others doing word processing and such. In a typical fat client set-up, you'd have to equip 100clients with a GB of memory, so that they are all suitable for all tasks (you do not want to have different configuration in a school, just one model for students, cause you will need spare systems, installations, servicing and purchasing procedures for each model you get). The server set-up just needs enough memory for the simultanously running apps. In other words, 100 fat clients here mean 100GB RAM, thin clients + server less then half of it.
Hi,
first, do not make any calculation based on buying machines, but on maintaining them. It maybe easy for you to get a budget for 100 laptops. But if these laptops are carried around by the kids, are to be used with current software, to be replaced every two years (max. life-span I would give to a mobile device under such conditions), you will need at least three additional stadd for doing just that. And how easy is it to get the budget for hiring staff? You do not want to have teachers, who should care about the kids, use their time in doing it administration and servicing, right?
I think my proposal would be a real thin client. A perfect solution may be something like Sun's Ray, as there is NO (zero) software installed on the client. This means that replacing a broken terminal requires you to plug-in one, that's it. Can be done by everyone who is able to change a light-bulb. Sun has huge reductions for educational institutions by the way. Attach this to a relieable server with some failover and such. Spend 2/3 of your it-budget in this, including its maintainance. Best is to get it with a service contract for the expected life-span. Do not even think about building this on your own or saving some dollars here, again, you do not want your teachers spend theit time on computers later.
While the work-hour of someone servicing a centralized server is usually higher then the salary of a guy offering you to take care of laptops for some dollars, there will be only one server to maintain. Every software installation will be done only once. You will not have kids with laptop bags waiting in queues to get their software updated. You can scale it up as needed (at least if you take a machine that is designed for this and not some PC in a server box). Add memory, CPU and such to the server when you need it, install the latest software, make backups - but NEVER have the it service and the kids in the same room;-)
The only reason for laptops is if kids really have to use it at home. But again, such a machine will be dead after two years, and replacement parts and servicing costs are high, so this will eat up your it and your human ressources budget.
Ok, we are back at the point where people who do not see a feature they need in blender start to invent a new app, often a rewrite a commercial one that we do not want to pay for. Here starts the trouble, so what is a parametric modeler? By definition it is creating models from parameters, thus changing parameters reflected in the model. You can do all kind of stuff with that idea, in 2d, 3d, on meshes, solids, generic objects, text... it does not define a cad yet.
I see two successful concepts of 3d applications. One is offering a great toolbox for one field of modeling, e.g. a great solid modeler, a cool 2d sketching CAD, a powerful mesh modeler. The user than uses tools from a pallette of apps, modeling solids in the modeler, drawing a plan in 2d cad etc, finally importing everything for rendering and output (rendering can be a raytracer or a 2d drafting/layout tool). The other way is not to write apps for the type of data that is worked on (solid, surface mesh etc) but the useage expected. As I understand that is how Blender developed, a tool for 3d animations. With that approach, the tool gets almost unuseable for anyone else of course, and here is the reason for most complaints about render. Do you want to make design (scale of mm, m, km?) requiring accuracy, simulation requiring simple but clean solid models, animation with high polygon count surface models and anim support, game design with low polygon count...?
Too many try to put everything for everyone's useage scenario into an application, that won't work. I think in open source the tool pallette concept works better, we have ways to do solid/csg stuff, surface meshes etc. The missing parts would be the solid cad package (that is what pro/engineer would refer to) and a really good 2d cad able to exchange data with the 3d foss tools.
No, communism is usually based on a centralized model (with few exceptions, in Jugoslavia something descentralized had been tried afaik, but at the times communication was a problem, long before internet;-). Of course, if you translate communism to public property, which it is in some meaning, then you are right.
I think the ideas is also not at all about some kind of anarchic structure with endless discussions. It is about keeping decisions closer to where they are going to be executed. So the argument that big groups mean slow decisions does not have to be true, as well as the argument that people would start to take discussions on fields they have no expert knowledge. An intelligent structure, something like a framework of groups, where one group shares work and decisions, could even be hierachical as there would have to be groups which take the results of others and assemble. In such a configuration, decisions could actually be much more precise and quick, as the delay of relaying information and decision through the hierarchy would be less limiting.
That is the basic about how open source works - we call the groups usually projects. If an application needs gtk to implement the gui, libc for standard calls, and a mathematic library to calculate, they won't start to take over the decision taking in all these projects. So the pretty thing about open source is that, while others build up on it, the project is still "owned" by those developing it.
The access to decision taking is usually simply controlled by how much someone contributes. So, in other words, if the marketing folks try to tell the engineer how to design a routine, they would not have much effect as they cannot contribute.
The question is if such a corporation would not be more like a network of small companies than one big one.
Lars.
"PC's are a large market serving many different types of users, yet they are first and foremost business machines. Customers of those systems demand and certify compatability. Apple systems aren't subjected to that kind of scrutiny."
Ok, I see. So the idea to have different machines for different specialised tasks being still compatible. This is hard to achieve with a typical Apple environment (OS X). Using Macs with Linux installations changes this, as Linux (with LSB and similar efforts) allow to have compatibility on virtually any platform (from cell phone to main frame). But of course, the typical Mac buyer never installs any OS on his machine, as he will just use the installed OS X as long as he uses the machine.
Of course, most Macs out there are business computers, too (actally there are few used for non-business stuff compared to the pc platform), but they target mainly two markets. One is unix folks, who like to have a nice supported unix system for affordable prices, the other is graphics in any meaning, where it is necessary to buy a complete system including machine, monitor and os e.g. for color display issues.
Lars.
I think you are the first I ever heard say the compatibility in the PC/Windows environment was superior to the Mac world...
At least in the past, Apple systems were rather famous not to suffer the problems caused by buggy and incomplete implementations of standards as they have been everyday for PC users. Maybe that is why PC users even do not realize these problems are there - for them it is often not a bug, but a technology;-)
How many years did it take until you could say that machines that were sold with some kind of acpi had a really compatible firmware, for example? I remember that we had to disable everything about it even in x86 server machines three years ago. At the time, in the speach of this thread, it was generation 4,5,6 products. Actually noone wondered. In the PC world, you do not get a note sent from the vendor that they had an implementation bug and fix it - instead you get a README that tells you that you either have to disable it or use a driver that works with the incompatible firmware.
So we Mac users are used to live in a world where instead of bios, every smallest mac used (not any more:-( openfirmware (as found in big servers usually) instead of some buggy and propietary "bios", taking it for granted that we could simply buy the newest OS and put it onto our Mac without downloading drivers, reading handbooks etc. The idea that compatibility is limited to new products should be set into a relation to the PC world, too... Actually, you can use the newest OS releases on more or less everything that has a G4 without reading through support webpages, downloading special drivers etc. If you take some effort, you will even find a way to install on older machines, as long as they have a PowerPC CPU. So the only architecture that really cannot be used for new OS X releases I am aware of are M68000-based Macs, and these are really old. Did you complain that it is hard to find a current OS to install on your x286?
I am not that convinced by a model where hardware and os are developed by the same. But at least it gives you a comfortable position as a user never facing compatibility problems. And as Apple machines have usually not had those undocumented design changes etc, they are even comfortable to support for other OS, e.g. Linux installations are usually really nice on a Mac.
So the only way I could understand your post is that you mean Microsoft drivers are compatible with the system, the hardware and the service-pack they were written for. And of course they are MS-Office compatible;-) Lets call them self-compatible... no standards.
Anything else in the PC area is wilderness, and that's why its fun, why we spend hours on patching, tweaking and reading the last fragment of documentation.
I am working here at the student's computer lab at the university of technology, Darmstadt. We have a lot of software in use, mainly on Mac and Windows-PC.
As we have a wide range of software (CAD, modelers, dtp, photoshop etc), we need a common base for all this, and so we only print postscript (ps, eps, pdf) and tiff, all in cmyk.
We have printed a lot on a large format postscript processor called Macrip, owned by Macron AG (www.macron-ag.de). Macrip runs on Solaris / Sparc and Linux / intel, in fact, it is ghostscript-based, and gives great results regarding speed, color and stability. It would be the perfect solution... but... at the moment, Macron sells it only on Windows... a "hack" running in a unix-environment on Windows NT..:-( In fact, that's what people buy at the moment. The current version is called Jack the Rip (www.jack-the-rip.com), it's quite an improvement regarding color, features and printer drivers.
Maybe there will be a Solaris version in future, one should simply ask... if they see there is enough interest, it would be almost no work to bring Ghostscript, Apache, Perl and their own things back into a real unix environment... (Linux, Solaris,...)
Slashdot Mirror
Yes, it is possible, while you usually do not want it. It is not easy to get a server system scalable for a whole school, with service contracts, warranty and stability for this application. The central application server requires hardware that fits pretty much into what Sun offers with their machines and solutions. If you are in a university, going with Linux here would be cool, as you would have lots of students to hire for working on this. But in a school, you do not want to have more trouble then saying "please get us 25% more RAM into our server and the latest software update" to the school secretary, so that she calls and gets the service installing this in less then a work-day...
The size of your Photoshop file should not matter, as it stays on the server. And you will have a server with a lot of memory. By the way, your use case is a great example - have 100 clients, and 40 of them running Photoshop, the others doing word processing and such. In a typical fat client set-up, you'd have to equip 100clients with a GB of memory, so that they are all suitable for all tasks (you do not want to have different configuration in a school, just one model for students, cause you will need spare systems, installations, servicing and purchasing procedures for each model you get). The server set-up just needs enough memory for the simultanously running apps. In other words, 100 fat clients here mean 100GB RAM, thin clients + server less then half of it.
Hi, first, do not make any calculation based on buying machines, but on maintaining them. It maybe easy for you to get a budget for 100 laptops. But if these laptops are carried around by the kids, are to be used with current software, to be replaced every two years (max. life-span I would give to a mobile device under such conditions), you will need at least three additional stadd for doing just that. And how easy is it to get the budget for hiring staff? You do not want to have teachers, who should care about the kids, use their time in doing it administration and servicing, right? I think my proposal would be a real thin client. A perfect solution may be something like Sun's Ray, as there is NO (zero) software installed on the client. This means that replacing a broken terminal requires you to plug-in one, that's it. Can be done by everyone who is able to change a light-bulb. Sun has huge reductions for educational institutions by the way. Attach this to a relieable server with some failover and such. Spend 2/3 of your it-budget in this, including its maintainance. Best is to get it with a service contract for the expected life-span. Do not even think about building this on your own or saving some dollars here, again, you do not want your teachers spend theit time on computers later. While the work-hour of someone servicing a centralized server is usually higher then the salary of a guy offering you to take care of laptops for some dollars, there will be only one server to maintain. Every software installation will be done only once. You will not have kids with laptop bags waiting in queues to get their software updated. You can scale it up as needed (at least if you take a machine that is designed for this and not some PC in a server box). Add memory, CPU and such to the server when you need it, install the latest software, make backups - but NEVER have the it service and the kids in the same room ;-)
The only reason for laptops is if kids really have to use it at home. But again, such a machine will be dead after two years, and replacement parts and servicing costs are high, so this will eat up your it and your human ressources budget.
Ok, we are back at the point where people who do not see a feature they need in blender start to invent a new app, often a rewrite a commercial one that we do not want to pay for. Here starts the trouble, so what is a parametric modeler? By definition it is creating models from parameters, thus changing parameters reflected in the model. You can do all kind of stuff with that idea, in 2d, 3d, on meshes, solids, generic objects, text... it does not define a cad yet. I see two successful concepts of 3d applications. One is offering a great toolbox for one field of modeling, e.g. a great solid modeler, a cool 2d sketching CAD, a powerful mesh modeler. The user than uses tools from a pallette of apps, modeling solids in the modeler, drawing a plan in 2d cad etc, finally importing everything for rendering and output (rendering can be a raytracer or a 2d drafting/layout tool). The other way is not to write apps for the type of data that is worked on (solid, surface mesh etc) but the useage expected. As I understand that is how Blender developed, a tool for 3d animations. With that approach, the tool gets almost unuseable for anyone else of course, and here is the reason for most complaints about render. Do you want to make design (scale of mm, m, km?) requiring accuracy, simulation requiring simple but clean solid models, animation with high polygon count surface models and anim support, game design with low polygon count...? Too many try to put everything for everyone's useage scenario into an application, that won't work. I think in open source the tool pallette concept works better, we have ways to do solid/csg stuff, surface meshes etc. The missing parts would be the solid cad package (that is what pro/engineer would refer to) and a really good 2d cad able to exchange data with the 3d foss tools.
No, communism is usually based on a centralized model (with few exceptions, in Jugoslavia something descentralized had been tried afaik, but at the times communication was a problem, long before internet ;-). Of course, if you translate communism to public property, which it is in some meaning, then you are right.
I think the ideas is also not at all about some kind of anarchic structure with endless discussions. It is about keeping decisions closer to where they are going to be executed. So the argument that big groups mean slow decisions does not have to be true, as well as the argument that people would start to take discussions on fields they have no expert knowledge. An intelligent structure, something like a framework of groups, where one group shares work and decisions, could even be hierachical as there would have to be groups which take the results of others and assemble. In such a configuration, decisions could actually be much more precise and quick, as the delay of relaying information and decision through the hierarchy would be less limiting.
That is the basic about how open source works - we call the groups usually projects. If an application needs gtk to implement the gui, libc for standard calls, and a mathematic library to calculate, they won't start to take over the decision taking in all these projects. So the pretty thing about open source is that, while others build up on it, the project is still "owned" by those developing it.
The access to decision taking is usually simply controlled by how much someone contributes. So, in other words, if the marketing folks try to tell the engineer how to design a routine, they would not have much effect as they cannot contribute.
The question is if such a corporation would not be more like a network of small companies than one big one.
Lars.
"PC's are a large market serving many different types of users, yet they are first and foremost business machines. Customers of those systems demand and certify compatability. Apple systems aren't subjected to that kind of scrutiny." Ok, I see. So the idea to have different machines for different specialised tasks being still compatible. This is hard to achieve with a typical Apple environment (OS X). Using Macs with Linux installations changes this, as Linux (with LSB and similar efforts) allow to have compatibility on virtually any platform (from cell phone to main frame). But of course, the typical Mac buyer never installs any OS on his machine, as he will just use the installed OS X as long as he uses the machine. Of course, most Macs out there are business computers, too (actally there are few used for non-business stuff compared to the pc platform), but they target mainly two markets. One is unix folks, who like to have a nice supported unix system for affordable prices, the other is graphics in any meaning, where it is necessary to buy a complete system including machine, monitor and os e.g. for color display issues. Lars.
Wow,
;-)
:-( openfirmware (as found in big servers usually) instead of some buggy and propietary "bios", taking it for granted that we could simply buy the newest OS and put it onto our Mac without downloading drivers, reading handbooks etc. The idea that compatibility is limited to new products should be set into a relation to the PC world, too... Actually, you can use the newest OS releases on more or less everything that has a G4 without reading through support webpages, downloading special drivers etc. If you take some effort, you will even find a way to install on older machines, as long as they have a PowerPC CPU. So the only architecture that really cannot be used for new OS X releases I am aware of are M68000-based Macs, and these are really old. Did you complain that it is hard to find a current OS to install on your x286?
;-) Lets call them self-compatible... no standards.
I think you are the first I ever heard say the compatibility in the PC/Windows environment was superior to the Mac world...
At least in the past, Apple systems were rather famous not to suffer the problems caused by buggy and incomplete implementations of standards as they have been everyday for PC users. Maybe that is why PC users even do not realize these problems are there - for them it is often not a bug, but a technology
How many years did it take until you could say that machines that were sold with some kind of acpi had a really compatible firmware, for example? I remember that we had to disable everything about it even in x86 server machines three years ago. At the time, in the speach of this thread, it was generation 4,5,6 products. Actually noone wondered. In the PC world, you do not get a note sent from the vendor that they had an implementation bug and fix it - instead you get a README that tells you that you either have to disable it or use a driver that works with the incompatible firmware.
So we Mac users are used to live in a world where instead of bios, every smallest mac used (not any more
I am not that convinced by a model where hardware and os are developed by the same. But at least it gives you a comfortable position as a user never facing compatibility problems. And as Apple machines have usually not had those undocumented design changes etc, they are even comfortable to support for other OS, e.g. Linux installations are usually really nice on a Mac.
So the only way I could understand your post is that you mean Microsoft drivers are compatible with the system, the hardware and the service-pack they were written for. And of course they are MS-Office compatible
Anything else in the PC area is wilderness, and that's why its fun, why we spend hours on patching, tweaking and reading the last fragment of documentation.
Hi!
:-( In fact, that's what people buy at the moment. The current version is called Jack the Rip (www.jack-the-rip.com), it's quite an improvement regarding color, features and printer drivers.
I am working here at the student's computer lab at the university of technology, Darmstadt. We have a lot of software in use, mainly on Mac and Windows-PC.
As we have a wide range of software (CAD, modelers, dtp, photoshop etc), we need a common base for all this, and so we only print postscript (ps, eps, pdf) and tiff, all in cmyk.
We have printed a lot on a large format postscript processor called Macrip, owned by Macron AG (www.macron-ag.de). Macrip runs on Solaris / Sparc and Linux / intel, in fact, it is ghostscript-based, and gives great results regarding speed, color and stability. It would be the perfect solution... but... at the moment, Macron sells it only on Windows... a "hack" running in a unix-environment on Windows NT..
Maybe there will be a Solaris version in future, one should simply ask... if they see there is enough interest, it would be almost no work to bring Ghostscript, Apache, Perl and their own things back into a real unix environment... (Linux, Solaris,...)
CU, Lars O. Grobe.