Ideas for RISC OS Plug ins... | |
(17:54 31/5/2000) The Doctor (19:04 31/5/2000) Matrix (17:09 1/6/2000) jess (08:46 13/6/2000) johnstlr (09:08 13/6/2000) jess (09:59 14/6/2000) ams (18:10 14/6/2000) Matrix (18:28 15/6/2000) Matrix (19:15 15/6/2000) Matrix (06:03 16/6/2000) jess (08:46 16/6/2000) johnstlr (15:20 16/6/2000) Matrix (18:01 16/6/2000) Matrix (18:13 4/7/2000) chume (11:30 16/6/2000) Matrix (17:21 1/6/2000) Matrix (17:25 1/6/2000) Matrix (11:30 2/6/2000) Matrix (09:08 6/6/2000) Matrix (14:13 7/6/2000) Matrix (14:17 7/6/2000) Matrix (14:23 7/6/2000) Matrix (14:26 7/6/2000) Matrix (14:30 7/6/2000) Matrix (11:27 8/6/2000) |
|
Matrix | Message #1275, posted at 17:54, 31/5/2000 |
Unregistered user | I think that risc os 4.xx is only one step far from RISC OS 3.7, i think that RISC OS must have this: 1) Full 32 bit system and you????? |
[ Log in to reply ] | |
The Doctor | Message #1276, posted at 19:04, 31/5/2000, in reply to message #1275 |
Unregistered user | I think that RiscOS 4 is a bigger improvement over 3.7 than Win98 was over Win95. It seems that many improvements are behind the scenes so to speak, and aren't immediatly apparent. However, one thing that RO4 is lacking and is definately needed is its own version of Windows 'Dial up Networking' This needs to be made as part of RO4 so as to set a sort of standard for the Platform. Currently we have Socketeer, Connect, and various others, none of which are a patch on Dial up Networking. A standard method of connection needs to be set, and I think it is up RiscOS Ltd to set it. Just my thoughts |
[ Log in to reply ] | |
Matrix | Message #1277, posted at 17:09, 1/6/2000, in reply to message #1276 |
Unregistered user | Good idea! A little explanation, i'm not ungry with RISC OS Ltd or other ARM company, but i really loveARM processor and i really hate Microsoft, and so i would like to see RISC OS like one of the best operating systems of the world and it CAN! |
[ Log in to reply ] | |
Matrix | Message #1278, posted at 17:21, 1/6/2000, in reply to message #1275 |
Unregistered user | Another good thing i think that can be a "RVD" Risc Virtual Device A meta language for pilot devices without know what real device it is, so you can make a program for work with all type of scanners etc... like printer driver that talk between you and the printer device... but for modems, interfaces, etc... |
[ Log in to reply ] | |
Matrix | Message #1279, posted at 17:25, 1/6/2000, in reply to message #1275 |
Unregistered user | Another one... Why RISC OS Ltd don't include a Graphics library like OpenGL (we have some Graphics library also professional) for example there is Be OS 5 have OpenGL Graphics library included in it! |
[ Log in to reply ] | |
Matrix | Message #1280, posted at 11:30, 2/6/2000, in reply to message #1279 |
Unregistered user | A very good idea from RISC OS Ltd is to include SAMBA Server a very powerfull Server plugin, thank you for this! :-) |
[ Log in to reply ] | |
Matrix | Message #1281, posted at 09:08, 6/6/2000, in reply to message #1280 |
Unregistered user | Well, is not hard to make a pool connection for RISC OS (also for RISC OS 4), THE PROBLEM IS MODEM DRIVERS... if RISC OS (5) will have a VDD (Virtual Device Driver) i think that a lot of peoples will write a lot of poll connections programs for RISC OS and RISC OS will select for you your right modem driver but if RISC OS continue to be a game OS and will not have the PROFESSIONAL FEATURES OF AN OPERATING SYSTEM it will never have a good Internet connection program (connection are not browsing!) |
[ Log in to reply ] | |
Matrix | Message #1282, posted at 14:13, 7/6/2000, in reply to message #1275 |
Unregistered user | I'm thinking that probably i'm the only one that want upgrade RISC OS to a new PROFESSIONAL OPERATING SYSTEM, well ok now i try to explain some RISC OS BIG problems (structural) so try to copy a program from one HD to another or from a CD ROM to your HD, in the same time try to DRAG one window on the screen without stop the mouse, you will see that the copy action WILL STOP!!! this type of multitask is NOT A REAL MULTITASK (this problem will be also if you use a SCSI and StrongARM and Kenetic and Imago AND WHAT YOU WANT) there is one other system that have the same problem WINDOWS 3.1 (for 386 processors!) well so also if you are not a programmer you undestand that RISC OS MULTITASK SYSTEM IS VERY OLD. So this si one BIG problem! |
[ Log in to reply ] | |
Matrix | Message #1283, posted at 14:17, 7/6/2000, in reply to message #1275 |
Unregistered user | How we can resolve this problem? Well some programmers did resolve this problems for all operating systems with PREMPTIVE MULTITASK SYSTEM and MULTITREADS but for use this system the operating system MUST USE PROTECTED MEMORY for don't let the process fight togeter when it re-take the machine control because a task are using too processor time. So RISC OS NEED PROTECTED MEMORY AND PREMPTIVE MULTITASK AND MULTITREADs. |
[ Log in to reply ] | |
Matrix | Message #1284, posted at 14:23, 7/6/2000, in reply to message #1275 |
Unregistered user | What is the cons of premptive multitask? ONLY A LITTLE LOSE OF Speed but because the system have to manage some VARIABLES and memory structure than the classic system... What is the pro? The pro is that your programs NEVER STOP because another program suck all processor resources because when it turn is end the system re-take controll and give resources to another application. |
[ Log in to reply ] | |
Matrix | Message #1285, posted at 14:26, 7/6/2000, in reply to message #1275 |
Unregistered user | What is the cons of protected memory? Protected memory need some memory buffers for save applications task datas and a different structure of resource memory. What is the pro? When a TASK CRASH the system and other tasks NO have problem or begin instable because all applications don't interact in real mode with system and hardware and other applications, so RISC OS WILL BE MORE STABLE than now... |
[ Log in to reply ] | |
Matrix | Message #1286, posted at 14:30, 7/6/2000, in reply to message #1275 |
Unregistered user | Why do you talk about VDM (Virtual Device Manager)? Well a V.D.M. is a program (like Printer manager) that be a shel between an application and REAL DEVICES that use a meta-language, so you can write an application, for example a connection-manager, and don't think about what modem have the end-user! so your connection-manager will work under all modem type, and if you want make a powerfull support you can use classic driver ways, but so you must try your program with all devices (very high cost) and make debug for all type of devices and for all your programs and forever! |
[ Log in to reply ] | |
Matrix | Message #1287, posted at 11:27, 8/6/2000, in reply to message #1286 |
Unregistered user | NEWS! Actualy i'm working arround WIMP2 Routines for make a good database of applications that can run in PRIEMPTIVE MULTITASK and maybe change some part of Wimp2 for make it better, there are someone that want know or receive the database with wimp2 version that run under RISC OS 4? |
[ Log in to reply ] | |
jess | Message #1288, posted at 08:46, 13/6/2000, in reply to message #1276 |
Unregistered user | I agree about an integrated Dial up networking system. I think this is the biggest omission from RISCOS when compared to Windows, especially I would think, from a non-techie user's point of view. Hopefully the same system would be made available for RO 3.x. All the features like memory protection, pre-emptive multitasking, multi-threading would be nice, but my RPC works far better than most Win PCs most of the time, without them. The only performance issue that annoys me is that the RO filing system seems slow compared to a PC. (Is this the multi-threading issue???) |
[ Log in to reply ] | |
johnstlr | Message #1289, posted at 09:08, 13/6/2000, in reply to message #1288 |
Unregistered user | I don't believe the fact that the RO filing system seems slow to you is really a multithreading issue unless each filing task is a complete task as opposed to a lightweight thread. Rather, if you think about how much Windows relies on it's filing system it's probably safe to say that it has been incredibly well optimised. I think most developers have agreed that memory protection should come first in order to improve stability even more. Pre-emptive multitasking and multi-threading offer a lot of benefits to developers as well. There has been some suggestion that pre-emptive multitasking will make application development more difficult but I think it will make it faster and easier. Firstly as a developer you no longer have to worry about monitoring your code and giving control back to the OS when doing long operations. Secondly multi-threading makes development of a lot of software - especially network based - far easier. However you do not have to use it if you are not confident with thread communication and synchronisation techniques. Multi-threading can also make applications more responsive. People have stated that downloading pages can lock a machine under RISC OS. This is easily avoided in a pre-emptively multitasked and multithreaded environment. In fact, in theory you could make RISC OS even more responsive than it is now. However it is true that both of these require the OS to do more work. This translates to a larger memory overhead and RISC OS being slower overall. If done properly the overheads will not be large whereas the benefits will be. |
[ Log in to reply ] | |
jess | Message #1290, posted at 09:59, 14/6/2000, in reply to message #1289 |
Unregistered user | I have heard a suggestion that a modified linux kernel, which has all these and security, should be used at the core of RISCOS. Might be nice. |
[ Log in to reply ] | |
ams | Message #1291, posted at 18:10, 14/6/2000, in reply to message #1290 |
Unregistered user | That might be throwing the baby out with the bathwater imho. True Linux has some really nifty features but its kernel and the RISC OS kernel are so different that it would be different to change ROS to a Linux kernel and still have all the existing code work. Besides who says the Linux kernel is the ONLY way to do things (that sort of thinking gave us Windows after all !). Anyway if a Linux kernel were grafted anything good that remains of RISC OS would be claimed (unjustifiably) by the Linux crowd :-) and they have enough support already ! A point of importance is RISC OS is inherently modular so there is no reason why it could not be upgraded in steps without hacking out the core (which after all is a major reason for RISC OS's speed). |
[ Log in to reply ] | |
Matrix | Message #1292, posted at 18:28, 15/6/2000, in reply to message #1291 |
Unregistered user | Good! so i think that is starting to be clear the the major new features of RISC OS must be: 1) Premprtive multitask I think that so we can start to image very powerfull RISC OS Servers and also (for example) good networks for computergraphics with new imago machines and some Risc OS server machines i think that this can be a wonderfull dream and can be a good idea for try to have programmers and developers, new good games, new server software, new professional applications for offices and industries... :-) |
[ Log in to reply ] | |
Matrix | Message #1293, posted at 19:15, 15/6/2000, in reply to message #1292 |
Unregistered user | Another one thing about a future of RISC OS in netrorking well if i use my Risc PC it is very easy configure it and control it so i think that it would receive a lot of consences in network server systems do you think the same???? |
[ Log in to reply ] | |
Matrix | Message #1294, posted at 06:03, 16/6/2000, in reply to message #1293 |
Unregistered user | Another thing (how many "another things Paolo!" :-)I think that if we need some more speed a full 32 bit system architecture is a good way, without change hardware.... |
[ Log in to reply ] | |
jess | Message #1295, posted at 08:46, 16/6/2000, in reply to message #1294 |
Unregistered user | What are the implications for compatibility of existing apps for Pre-emptive multi-taking, Memory protection, Multi-threading and 32 bit? If it gets a major re-write, Linux\Unix style security should be included. |
[ Log in to reply ] | |
chume | Message #1296, posted at 11:30, 16/6/2000, in reply to message #1291 |
Unregistered user | Also depends what you use your computer for, I've always thought that Acorn has a relativley large graphic design section of it's community. Linux's kernel is one of the worst around for graphics. |
[ Log in to reply ] | |
johnstlr | Message #1297, posted at 15:20, 16/6/2000, in reply to message #1295 |
Unregistered user | Memory protection would have the benefit that apps would not be able to trash each others workspace. This would make current apps more stable. Multithreading will not have any effect as applications have to be written to exploit it. Pre-emptive multitasking should make it impossible for a misbehaving app from hogging the processor. 32 bit processors (ie ARM8+) will not run current apps. If the app is written in C / C++ then a simple recompile will solve it. ARM code will require a rewrite / patch or somekind of emulation as the standard way of entering and leaving a subroutine in ARM is not guaranteed to preserve the processor flags on 32bit hardware. Unfortunately pretty much all code assumes that flags are preserved. |
[ Log in to reply ] | |
Matrix | Message #1298, posted at 18:01, 16/6/2000, in reply to message #1297 |
Unregistered user | Sounds very good! ok now i want say another thing... a lot of peoples know that ARM is the only one (low price) processor that be FAST and low electrical use, this mean that a NETWORK SERVER based on ARM Architecture can be fast and can work more than a pentium based server this mean major security in circuit work and also major work time, if the RISC OS will be the O.S. type that we want well future machine (with a most fast HD acces than now) will be another way for ARM processors not only graphics and in this mode we can image full ARM networks based! more powerfull than pentium based network and also web servers based on ARM will be more fast than pentium based servers, actualy we can use also parallel architecture and can image a new motherboard fast with 2 StrongARM? .... |
[ Log in to reply ] | |
Matrix | Message #1299, posted at 18:13, 4/7/2000, in reply to message #1298 |
Unregistered user | Why not implement the possibility to use another SA in parallel??? maybe will be program caching problems but i think that it is possible..... 2 SA and RISC OS will be a perfect game machine and also a Server machine for professional applications.... |
[ Log in to reply ] | |