On Tuesday 24 October 2006 17:49, Markus Gritsch wrote: > Since Python string objects are immutable, a new instance of the > object has to be created, which can become expensive, if the strings > become quite long. A performant way to concatenate Strings is to add > the strings to a list, and join the list at the end. Executing the > following script on my computer using Python 2.4 for 100000 iterations > takes 0.063 vs. 0.062 seconds, so no big difference, but when > performing 1000000 iterations the difference becomes quite large: > 6.125 vs. 0.532 seconds, so joining the list is more than 10 times > faster. Things get worse for string concatenation with larger > strings.
The discussion and comparison was between string concatenation and the % operator (which apply to creating strings from a small number of other elements), not the join method which has other semantics (combining many strings together). I doubt that anyone would try to concatenate 1000000 strings by appending them one to another (well, I wouldn't actually bet on this ;) but you got my point), or by trying to use the % operator. But since you mentioned it, I have obtained different results by running your test: dawn:~$ python2.4 a.py 1.04168891907 0.631967067719 which means that string concatenation was only 40% slower, not 10 times slower for me. Now considering the way this test was setup I'd say it is extremely biased. The 1st method creates/destroys 1000000 string objects, while the 2nd only creates 1 at the end. Also the 2nd method doesn't even add new objects to the list, it keeps appending a reference to the same string, which means in the second case you only have 3 objects overall: the list, the 'hello' string and the result string. Given these conditions, I'm amazed that the 1st case was only 40% slower. If you really want to compare them 1:1 you should execute the same number of operations overall (i.e. as many joins as concatenations). -- Dan ------------------------------------------------------------------------- Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 _______________________________________________ sqlobject-discuss mailing list sqlobject-discuss@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/sqlobject-discuss
