Re: [xml] Minor patch

On Wed, May 01, 2002 at 03:44:52PM -0400, Daniel Veillard wrote:
The attached patch should provide more sensible behaviour in this
pathological case -- we only grow the hash table if the *average* chain
length is greater then MAX_HASH_LEN (perhaps MAX_HASH_LEN should also be
reduced if this patch is applied?)
  Hum, While I understand the specific case, I really wonder what is the
associated penalty in term of speed for the whole class of applications.
For example timing impact on DocBook XSLT transformations, or the
XSLTMark result before and after.
  I'm not tempted to apply this patch until I have some report on this.

Fair enough, I'm afraid I don't have any performance test data, and as it's
a problem that's rather unlikely to crop up in real life I suppose avoiding
the change is the safer option (although I do suspect that that the
performance penalty would be fairly minimal, especially if MAX_HASH_LEN is
reduced in conjunction with applying the patch)...



[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]