 |
 | Home | Contact Us | FAQ | Search & Site Map | Link to Us |
 | Sign In | Join | Other 45 Sites in Network |
Java Forum / General / January 2006opting out of Iterators
I am constantly trying to write my code (regardless of language) in the lowest common denominator syntax (I'm too old, I've had to rewrite too much code in my days! By doing this, conversion of code segments to other similar languages in the future is far simpler). That is, code that essentially looks the same in MOST languages (c, c++, php, java, c#). Thus, I try to avoid, templates, generics, etc. As such, I also try to avoid Iterators in Java. Say I have ArrayList arrayList; //will hold String variables Then I will "iterate," through this as: for(int x = 0 ; x< arrayList.size(); x++){ String s = (String)arrayList.get(x); // ... } Yes, I know that J2SE1.5 allows generics and such (of which I prefer not to use for the same reasons mentioned earlier) however my real question here is, since this seemingly more draconian means of iterating through a collection, as demonstrated here, does, in fact, isuccefully iterate through just as an Iterator does, am I suffering a sufficient performance hit by doing so? TIA, Ike [...] > As such, I also try to avoid Iterators in Java. Say I have > [quoted text clipped - 13 lines] > just as an Iterator does, am I suffering a sufficient performance hit by > doing so? As long as you are using an ArrayList, there really isn't much of a reason not to use this style (other then the convenience of the maintainance programmer, that is!), but what if you do not use an ArrayList, but a more general List, without really knowing the implementation? It might blow the costs for the loop from n to n². Also, you lose on type safety by using the raw list type rather then the generified version. If this is an acceptable tradeoff for having terser, "more portable" (as in easily rewriteable) code... i am not going to tell you you are wrong, but i would put the priorities the other way around. > [...] > [quoted text clipped - 21 lines] > ArrayList, but a more general List, without really knowing the > implementation? It might blow the costs for the loop from n to n². Since the iterator on a particular List instance is returned as an interface, it seems implied that it does more than you expect. Not only for loop performance on iteration alone, but (for example) I've dealt with smart pointer libraries in C++ where it was smart iterators and iteration through the list that cleaned up dead references. In applications where the iterator does more than just dereference and move it becomes more important to use them rather than the loop-counter increment/grab way. --Paul > Since the iterator on a particular List instance is returned as an > interface, it seems implied that it does more than you expect. Not only [quoted text clipped - 6 lines] > > --Paul Do you - or anyone who may happen upon this message, know if there is anything else performed by an Iterator in java like you mention here, which may not meet the eye? Thanks, Ike >>Since the iterator on a particular List instance is returned as an >>interface, it seems implied that it does more than you expect. Not only [quoted text clipped - 4 lines] >> it becomes more important to use them rather than the loop-counter >>increment/grab way. > Do you - or anyone who may happen upon this message, know if there is > anything else performed by an Iterator in java like you mention here, which > may not meet the eye? Thanks, Ike I do not think that the iterators of Java's standard List implementations do anything similar to the behavior Stefan described. Moreover, I think it broken indeed that any iterator, anywhere, does that kind of work -- "more than you expect" is never what you want of a component. I would consider the particular C++ library Stefan mentioned to be quite buggy, unless the smart iterators he mentioned were the *only* way to access list elements. On the other hand, people can and do write shockingly bad Java code. (And C++ code just as bad, but it's less shocking :-)) It is entirely possible that some custom List implementation you run into one day might play tricks with iterators. It is up to you whether or not to worry about it.  SignatureJohn Bollinger jobollin@indiana.edu > >>Since the iterator on a particular List instance is returned as an > >>interface, it seems implied that it does more than you expect. Not only [quoted text clipped - 16 lines] > to be quite buggy, unless the smart iterators he mentioned were the > *only* way to access list elements. You fell for a misquote, i did not write what is attributed to me. > I do not think that the iterators of Java's standard List > implementations do anything similar to the behavior Stefan described. > Moreover, I think it broken indeed that any iterator, anywhere, does > that kind of work -- "more than you expect" is never what you want of a > component. Indeed, but only when it breaks the abstraction. If the "more then expected" were to rebalance a balanced tree (assuming, for some odd reason, that the rebalancing done by Iterator were net beneficial for performance reasons over that done by get(int)), then I wouldn't see a problem. This falls into the same category as the LinkedList issue, of course. Signaturewww.designacourse.com The Easiest Way To Train Anyone... Anywhere. Chris Smith - Lead Software Developer/Technical Trainer MindIQ Corporation >Do you - or anyone who may happen upon this message, know if there is >anything else performed by an Iterator in java like you mention here, which >may not meet the eye? Thanks, Ike If you changed your code from an ArrayList to any other collection, the code that iterates through the elements is unchanged. SignatureCanadian Mind Products, Roedy Green. http://mindprod.com Java custom programming, consulting and coaching. > Do you - or anyone who may happen upon this message, know if there is > anything else performed by an Iterator in java like you mention here, which > may not meet the eye? Thanks, Ike The extra thing Java iterators do as standard is to make some kind of a "best effort" attempt to detect concurrent modifications (not necessarily through a different thread). java.util.concurrent CopyOnWriteArrayList/CopyOnWriteArraySet work on a consistent copy of the collection. Other iterators in the same package have other peculiar contracts. Tom Hawtin SignatureUnemployed English Java programmer http://jroller.com/page/tackline/ > I am constantly trying to write my code (regardless of language) in the > lowest common denominator syntax (I'm too old, I've had to rewrite too much I was feeling far too old in the Dartmouth Inn two nights ago. Now all of a sudden I feel all young again. > code in my days! By doing this, conversion of code segments to other similar > languages in the future is far simpler). That is, code that essentially [quoted text clipped - 11 lines] > // ... > } Sure you don't want to replace that with?: for (String string : list) { // ... } Okay, but in C and C++ you would write something completely different from either of those. > Yes, I know that J2SE1.5 allows generics and such (of which I prefer not to > use for the same reasons mentioned earlier) however my real question here > is, since this seemingly more draconian means of iterating through a > collection, as demonstrated here, does, in fact, isuccefully iterate through > just as an Iterator does, am I suffering a sufficient performance hit by > doing so? For ArrayList you will probably go slightly faster using a loop counter rather than an iterator. For small lists, you are doing one less very small allocation. For large lists, there is one less level of indirection. When I last tested sometime ago, server/C2 HotSpot mostly removed the iterator penalty. Client HotSpot was back then significantly slower, if I had a negligible loop body. If there is a performance problem, one of my first moves would be to move the loop into its own private method. HotSpot avoids inlining into large methods, which can cause problems. Tom Hawtin SignatureUnemployed English Java programmer http://jroller.com/page/tackline/ >for(int x = 0 ; x< arrayList.size(); x++){ > String s = (String)arrayList.get(x); > // ... On the other paw, there is a lot less visual clutter with: for ( String s : arrayList ) { System.out.println( s ); } The less clutter the lower the odds a typo will bite you. e.g. >for(int x = 0 ; x< =arrayList.size(); x++){ > String s = (String)arrayList.get(x); > // ... >for(int x = 0 ; x< arrayList.size(); x++){ > String s = arrayList.get(i); > // ... >for(String x = 0 ; x< =arrayList.size(); x++){ > String s = (String)arrayList.get(x); > // ... SignatureCanadian Mind Products, Roedy Green. http://mindprod.com Java custom programming, consulting and coaching. > I am constantly trying to write my code (regardless of language) in the > lowest common denominator syntax (I'm too old, I've had to rewrite too much [quoted text clipped - 13 lines] > // ... > } int i = 0; while (i < yourArrayList.size()) { String s = (String) yourArrayList.get(i); i++; // not tested } > Yes, I know that J2SE1.5 allows generics and such (of which I prefer not to > use for the same reasons mentioned earlier) however my real question here [quoted text clipped - 4 lines] > > TIA, Ike SignatureNoodles Jefferson mhm31x9 Smeeter#29 WSD#30 sTaRShInE_mOOnBeAm aT HoTmAil dOt CoM NP: "The Road to Chicago" -- Thomas Newman (Road to Perdition Soundtrack) "Our earth is degenerate in these latter days, bribery and corruption are common, children no longer obey their parents and the end of the world is evidently approaching." --Assyrian clay tablet 2800 B.C. Ike sez: ... > for(int x = 0 ; x< arrayList.size(); x++){ > String s = (String)arrayList.get(x); [quoted text clipped - 3 lines] >... am I suffering a sufficient performance hit by > doing so? No. Potential performance hit is where you resize an ArrayList: that requires twice the memory. If you run into that and get the bright idea of fixing it by using LinkedList instead, your loop will suddently go from O(n) to O(n^2). Dima SignatureRiding roughshod over some little used trifle like the English language is not a big deal to an important technology innovator like Microsoft. They did just that by naming a major project dot-Net (".Net"). Before that, a period followed by a capital letter was used to mark a sentence boundary. --T. Gottfried, RISKS 21.91 > I am constantly trying to write my code (regardless of language) in the > lowest common denominator syntax (I'm too old, I've had to rewrite too much [quoted text clipped - 13 lines] > // ... > } If the list were simply that - an instance of a class that implemented java.util.List, you might (or might not) get very poor performance. If the List were an instance of java.util.LinkList the performance would be horrid, since the get(int) method for a linked list is NOT constant time, it's O(N), which means your loop would be O(N^2) Further, one can have iterators over things which don't even have a get(int) method. I use this a lot, since the jakarta Collections package allows me to do such scrummy things as take the Iterators for 2 objects (e.g. your ArrayLists) and scrunch them together to make an Iterator that traverses the 2 Lists consecutively. The point is, the concatenated object that the Iterator is traversing doesn't even exist to have a get() performed on it... http://jakarta.apache.org/commons/collections/api-release/org/apache/commons/col lections/iterators/IteratorChain.html BugBear BugBear > I am constantly trying to write my code (regardless of language) in the > lowest common denominator syntax (I'm too old, I've had to rewrite too > much code in my days! By doing this, conversion of code segments to other > similar languages in the future is far simpler). Presumably, as you get older, the number of re-writes you can anticipate in the time remaining to you will drop towards zero. Also (presumably) the value you place on what little time you have left will grow. Someday the two curves will cross, and you'll be able to write normal code like a normal programmer. I would look forward to that day, if I were you. I would even anticipate it. -- chris Free MagazinesGet these publications absolutely FREE for up to 12 months. There are no hidden fees and no obligation. Simply choose a title, complete the application form and submit it. Read more ...
|
Reply to this Message
|
 Sign In
 Join
 My Latest Posts My Monitored Threads My Blog My Photo Gallery My Profile My Homepage Start New Thread Enable EMail Alerts Rate this Thread
|
©2008 Advenet LLC Privacy Policy - Terms of Use
This website includes both content owned or controlled by Advenet as well as content owned or controlled by third parties.