Talk:Semaphore (programming)
From Wikipedia, the free encyclopedia
 |
This article may be too technical for a general audience. Please help improve this article by providing more context and better explanations of technical details to make it more accessible, without removing technical details. See below for more information. |
Contents |
[edit] Assembly language instruction
Can anyone pls tell me which assembly language instruction supports construction of semaphores ? - Computer Freak
- @computer freak. There is no assembly instruction. You need an instruction to dis-/enable interrupts. Then you can make 'atomic' pieces of code.
- The P is short for 'Prolaag' which means try to decrement. The V is short for Verhoog which means increment. Please check it in the original document: http://www.cs.utexas.edu/users/EWD/ewd00xx/EWD74.PDF
- Lex Meuldijk
-
- @Lex. "Prolaag" doesn't mean anything -- it is not a dutch word. It is probably a shorthand for 'proberen te verlagen' which means 'try to decrement'. Which would make sense because the other operation is indeed "verhogen", which means "increment".
-
- Jan David Mol
[edit] Supporting information
Could anyone provide supporting information for the statements in the "semaphores today" section?
I fail to understand why semaphores should be considered as bad as goto.
The only info I was able to google did not clarify it (such as: preferring the synchronize keyword for the Java language, the difficulty of the use of semaphore).
Thanks!
--Fr3d3r1c 17:50, 27 Feb 2005 (UTC)
Actually the text does not say semaphores are as bad as GOTO. The comparison between semaphores and GOTO is that both seemed to be good ideas in the beginning. A personal view is that a semaphore is just the simplest implementation which does the job and employs 'atomic' or 'atomicised' instructions. More modern alternatives will do the same job by the same underlying means, but will wrap them up differently for easier application to a programming problem. Much like a roller makes it possible to move heavy loads without sliding friction and a wheel is a better implementation of the same idea, because you don't need to keep moving rollers from behind to the front.
VL
[edit] I don't understand the example
I'm sorry to ask stupid questions, but I don't fully understand the example:
...It (thread A) then posts a DBDataRequest to both the threads(B,C) and adds a reference to the semaphore as well in the request it posted. After this it immediately does P(S) and blocks. The other two threads meanwhile take their own time to obtain the information and post the responses back and also do a V(S) on the passed semaphore. Only after both threads have done their part of V(S) will the first thread be able to continue. This is exactly what we wanted as well. A semaphore used in this way is called a "counting semaphore".
If the pseudocode in thread A is something like
read data from B read data from C P(S)
then won't it proceed as soon as either B or C has done V(S)? Should there be two calls to P(S)? I'm reading the example text like there is only one call to P(S)... ~Samuli
- Because there is a problem with the example. The call to "init(S,0)" SHOULD BE "init(S, -1)".
- Thread A calls P(S), which will hang, or wait, until the semaphore is > 0. And the semaphore will only be > 0 when there have been at least 2 V(S) calls.
- Hope this helps. - Jeremiah
[edit] C# example
I would like to get rid of the C# example. My feeling is that it is way too long for an illustrative source snippet and makes the article unreadable. Any comments? Abelsson 10:14, 23 January 2006 (UTC)
- I agree. -dkeithley
-
- As there were no objections in over a week, I've removed it now. Abelsson 08:49, 1 February 2006 (UTC)
[edit] Article "Design patterns for semaphores" needs ACM membership
Shouldnt the above article be excluded as it is not freely availble?
[edit] Try-and-decrease
Yes, Dijkstra really wrote "try-and-decrease," not "try-to-decrease." That's an idiomatic English expression; for example, "I'm going to try and improve this article a bit." However, "try-to-decrease" is a much less ambiguous phrase. (A non-native English speaker, or even a paranoid native speaker, might sensibly ask, "Try what, and (then) decrease?") So I've made the change in the article. I don't think it's appropriate to keep only the "try-to-decrease" translation, since that's not what Dijkstra actually wrote. --Quuxplusone 07:18, 3 March 2006 (UTC)
@Quuxplusone:
Where did Dijkstra write "try-and-decrease"? He wrote all his papers in Dutch. If you got it from http://www.cs.utexas.edu/users/EWD/transcriptions/EWD00xx/EWD51.html, I think the translations were added by the transcriber, they do not appear in the original document: http://www.cs.utexas.edu/users/EWD/ewd00xx/EWD51.PDF
[edit] deadlocks
Someone should mention why semaphores can't handle deadlocks: because a random thread gets unblocked when V is called (so the threads don't wake up in FIFO order; it is theoretically possible that a thread never wakes up if other threads call P all the time and get woken up first). --Bernard François 17:39, 4 January 2007 (UTC)
