On Jan 9, 2006, at 22:20, [EMAIL PROTECTED] wrote:

Author: adelmelle
Date: Mon Jan  9 13:20:07 2006
New Revision: 367395

URL: http://svn.apache.org/viewcvs?rev=367395&view=rev

OK folks,

This should solve the 'trailing white-space for trailing (nested) inlines' issue recently discussed ad nauseam here :-)

I was a bit doubtful at first about using mark() for this, since I received a couple of concurrent modification errors. Those turned out to be caused by blocks being added to the pendingInlines list, so concurrent modification of the underlying list of white-space was to be expected. Once this was fixed, all regression tests passed w/o complications. It should work since *iff* there is remaining white- space (inWhiteSpace==true), then the element at which the iterator was marked, is guaranteed to be present (and the rest of the list unchanged if white-space-collapse="false").

Again, no noticeable change in the iterators yet. Strictly speaking, at least the InlineCharIterator could already be removed, I think (not used anymore), but I consider the iterators a task in itself, so will take care of it then and there.

Markers, as Manuel suggested, are currently still white-space-treated according to their possibly erroneous property values, since that would still cover most of the use-cases. Obvious work-around is to use a block simply as a container for the related properties to be inherited by the marker-content.



Reply via email to