| ??? 02/22/05 20:41 Read: times |
#88184 - One task or three Responding to: ???'s previous message |
Erik Malund said:
yes, I disagree, the "task" is "print". The example you have given is a dot matrix printer controller and if you separate these 3 functions you will have no guarantee that "prepared print data" will be ready when the motor has located the printhead to a given position. Likewiase, if you separate tne functions you have no guarantee that the printhead is located where you want to be when you control the print head pulse width. So, as all depend on all, you can not split it. This is simply not true. The fact they depend upon each other does not mean they cannot be considered separately. In fact they are separate because two of them take place in different ISRs and the third runs in the background. I am, of course, aware that you can make 'a' set a flag that 'b' wait for etc. and you can, of course design a printer where the motor stop and wait for "prepare print data", then "control the print head pulse width", then start the motor and repeat.
I doubt you would sell many. I agree. This is a very poor design but has nothing to do with what we are discussing. The design I described has sold in the millions. The issue is that if the functions depend on each other, you can not separate them into "tasks". The fact they need to communicate with each other makes no difference to whether thay can be considered tasks. The example of Windoew where no "task" depend on the other, your ICE run fine even if you have IE running BECAUSE they do not depend on each other. Which is completely different. That is several applications running on a PC. We are discussing a single application consisting of several tasks. The point is, I don't really care what you call them. The point is it is usefull to consider them this way in the design process because it emphasises the need to handle the communication between tasks. I have seen far too many applicaions using a simple workloop that fail to realise this and end up with the nightmare of the occasional random failure. In one way I am glad lots of people still code this way as it provides lots of work for me. Ian |
| Topic | Author | Date |
| please expalin RTOS, UnRTOS, MTOS... | 01/01/70 00:00 | |
| Os's | 01/01/70 00:00 | |
| thanks ...and now for Erik... | 01/01/70 00:00 | |
| MicroC O/S II works fine | 01/01/70 00:00 | |
| Threads? | 01/01/70 00:00 | |
| Threads, Real Time and Multitasking | 01/01/70 00:00 | |
| Just for clarification | 01/01/70 00:00 | |
| Multiprocessing | 01/01/70 00:00 | |
| Multiprocessing | 01/01/70 00:00 | |
| Further clarification | 01/01/70 00:00 | |
| Threads | 01/01/70 00:00 | |
| Windoze | 01/01/70 00:00 | |
| More Windoze | 01/01/70 00:00 | |
| Ah... | 01/01/70 00:00 | |
| Basic Info | 01/01/70 00:00 | |
| extract from the above with translation | 01/01/70 00:00 | |
| Completed | 01/01/70 00:00 | |
| who are we | 01/01/70 00:00 | |
Tics | 01/01/70 00:00 | |
| the rest | 01/01/70 00:00 | |
| But wait..there's more! | 01/01/70 00:00 | |
| Real Time Info | 01/01/70 00:00 | |
| hard/soft | 01/01/70 00:00 | |
| Real Time Info | 01/01/70 00:00 | |
| m,y opinion | 01/01/70 00:00 | |
| why so late | 01/01/70 00:00 | |
| one more thing | 01/01/70 00:00 | |
| Architecture Independent | 01/01/70 00:00 | |
| just to clarify | 01/01/70 00:00 | |
| No. | 01/01/70 00:00 | |
| RTOS on a 51?? | 01/01/70 00:00 | |
| not the thing but the word | 01/01/70 00:00 | |
| A Distinction (with a difference) | 01/01/70 00:00 | |
| Tasks or Applications? | 01/01/70 00:00 | |
| Multi Tasking | 01/01/70 00:00 | |
| a dot matrix printer must be one task | 01/01/70 00:00 | |
| One task or three | 01/01/70 00:00 | |
| pray explain | 01/01/70 00:00 | |
| Tasks and Functions | 01/01/70 00:00 | |
| we are getting closer | 01/01/70 00:00 | |
| Not that far apart | 01/01/70 00:00 | |
| I doubt we are | 01/01/70 00:00 | |
| discussed | 01/01/70 00:00 | |
| TTCS | 01/01/70 00:00 |