Tcl is often somewhat slower than Perl for comparable operations. Algorithms that match Perl's regular expressions well generally favor Perl, sometimes by a large factor; a lot of work has gone into optimizing Perl's REs. On the other hand, Perl REs sometimes are only a thousandth as fast as Tcl's, and, in one very public case [1], Tcl is over four times as fast as Perl on a comparable RE chore. Lambda-the-ultimate briefly discusses how this can be true [2].
[Explain other penalties--correct Unicode and thread handling, ...]
Naive Tcl coders rarely understand the simple steps that boost Tcl performance. [Explain why this affects Tcl differentially.]
See http://www.equi4.com/md5/
for one example where Tcl is not that much slower.Perhaps JCW will update these tests to reflect the new md5 code...Arjen Markus From my own experience I can add that Perl and Tcl may have different ideas about what seems the same task. For instance: reading a file line by line. I noticed in a slightly naive test that Tcl would convert the end-of-line sequence, and Perl would not. (Note: I have not delved into the details and it may be specific to the particular releases I used then). This means that setting up a performance test requires a thorough understanding of what each language will do.Furthermore, slight changes in the implementation of the algorithms may make all the difference. (Somewhere I see apples and pears lurking behind shrubberies of intricacies!)
DKF - Just noting here that 8.4 is the fastest Tcl to date, even faster than 8.0, as listed in Tcl Benchmarks. Despite having i18n and thread support...
Setok - The topic isn't even true for straightforward examples. I converted the Perl intro available at http://www.altparty.org/archive/invi.pl
direct to Tcl 8.4 and it was faster (there is also a more optimised version at http://www.altparty.org/archive/invi.tcl).Tcl's not that slow compared to Perl: http://shootout.alioth.debian.org/gp4/benchmark.php?test=all&lang=tcl&lang2=perl
