On Tue, 2003-02-11 at 16:50, Roger Horne wrote: > On Tue 11 Feb, Dave Hodgkinson wrote: > > > > Hrm, perl doesn't use UTF-8 for it's internal representation of strings ? > > > > I may be misremembering things somewhat. The cloudy things I remember > > are that the version of perl shipped with RH8 uses a slightly geb0rken > > 16 bit implementation (who is the guy who is developing it? ISTR he's > > thrown out or totally changed how it works) the cripples performance. If > > you unset LANG perl goes a lot quicker. > > I must try that. I am an amateur and I use an overly complicated and badly > written script to process legal material, turning it from basic HTML and > text into fully marked-up HTML. I used to run it on an Acorn RPC using > Nicholas Clark's port of Perl but it started taking longer and longer to run > (because there was much more material). Eventually it used to take 5 hours. > > So I bought a fastish machine from Digital Networks last year and the time > was reduced to 5 minutes on RH 7.2. > > I then tried it on a reasonably fast netbook on which I had put RH8. After > 15 hours I gave up on it ... Possibly not enough memory (although more than > the Acorn) and the machine is going back to DN to correct another problem, > but it did seem excessive. (My main machine is still running RH 7.2 as a > result.) >
Methinks there is much smelly fish in Perl 5.8.0 UTF implementation and also in RH8.0's internationalisation stuff / port of perl. A standard 'cure' seems to be to rename /etc/sysconfig/i18n to something else (eg i18n.orig) and then restart the machine. Suddenly things like man pages start to come out in something approaching english (ie without a load accented a's instead of <tab> and stuff) and your perl goes a lot faster. Quite what you do if your language isn't accommodated by that - I just don't know. Please don't be tempted to use the -d switch with 5.8.0 because otherwise it will also run about 18 times slower... (sorry to harp on about it chaps but this is sooooo fundemental I am surprised it took me to pick it up several months after release). Dirk -- Please Note: Some Quantum Physics Theories Suggest That When the Consumer Is Not Directly Observing This Product, It May Cease to Exist or Will Exist Only in a Vague and Undetermined State.