On Tuesday 31 December 2002 00:20, "Lars Steiger" wrote: > do you address the implementation of the "shouldSetContentLength true case" > inside CachingStreamPipeline ? or why do you mean, that it can cause memory > problems and cache congestion.
Its a matter of the serializer only. If "shouldSetContentLength" is true, the whole result is written into a memory byte stream in order to get the content length which must be sent before the content (in a HTTP header, you know). Because PDF results can be quite large, this takes up quite some memory, and the repeated buffer expansion for the stream can cause additional problems due to memory fragmentation. If no content length has to be set, the result could be streamed to the client directly without much buffer reallocation and such. As for the possible cache congestion: if the large PDF result is cached, it may throw lots of small but often used items out of the cache, like GIF icons. I'm not sure how caching is handled if the result is directly streamed to the client, or if this changes once "shouldSetContentLength" is true and the result is sent as a single block. Is one of the developers reading this? J.Pietschmann --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, email: [EMAIL PROTECTED]
