On 3/3/06, Will Glass-Husain <[EMAIL PROTECTED]> wrote: > Hi, > > I'd argue that SimplePool is a utility class and not part of the public API > (so it's ok to change this). Have to draw the line somewhere - otherwise we > can never change anything. Anyone object? > > I'm not really following the problem with factories -- I'll wait to see > until I see some code to comment on this. > > Llewellyn Falco has a Eclipse code formatting file. I had a copy but my > computer is in the shop. Llewellyn are you out there? > > Best, WILL
I reran my benchmarks to conclude whether parser pooling is necessary. I changed to 0 parsers in the pool (so a new one is created each template compilation) for these numbers... 4k template file 3.4Ghz HP laptop Win JDK 1.5 run 1000 times 5,250ms to compile template by reusing parser 6,953ms to compile template creating a parser each time So given 1,000 iterations, about 1.7ms to create a parser. WRT my comments about refactoring, the SimplePool is IMO not a pool but an array. Right now, RuntimeInstance creates simplepool, iterates through the # of spots in the pool, constructs a parser and assigns it to that spot. Then whenever RuntimeInstance needs a parser, it checks if one available in simplepool, and if not, it constructs a new one. I think of a pool as providing object instances and the user does not know whether it is a reused or new instance. To do this, a pool needs a factory to create a new one when necessary. Hope that clarifies. -- Serge Knystautas Lokitech >> software . strategy . design >> http://www.lokitech.com p. 301.656.5501 e. [EMAIL PROTECTED] --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
