2006/2/3, Jochen Wiedmann <[EMAIL PROTECTED]>: > On 2/3/06, Henri Gomez <[EMAIL PROTECTED]> wrote: > > > Using filesystem as cache could be a good idea but it may > > introduce performance penalties. > > That's why it has to be configurable and by default turned off. > > > BTW, it could be proposed as > > extension to those who want minimal memory footprint. > > Not only those. Take, for example, a 30MB text file. In memory, this > may easily take 60 or 100MB (remember character encodings). That's far > away from "low memory".
:) > > bytes arrays or strings of Java Objects to be stored in the file cache > > ? And at a latter time transform them into XML stream ? Did the two > > passes won't introduce again some penalties ? > > I am currently talking about the parser only. Streaming on the writing > side is easy: You simply add support for InputStreams or Readers to > the TypeFactory. (May well be it already is there.) ok should take a look at this > On the parsing side, the main problem is to decide on whether memory > or disk shall be used. My current proposal uses the size for a > decision. A better algorithm might be to look at the target methods > parameters. Well everything depend on the volumetry, if the data to parse are huge, it may be a good reason to cache on disk.
