6 commits

Author	SHA1	Message	Date
jez	fe40d482f5	Aha! The push/pop precedence thing is, of course, wrongly wrong. Need to assign the new precedence at the point we encounter the xsl:import, then set that precedence as current when we actually load it. I'll sort that out next time, because it's bedtime for programmers now.	2008-11-24 23:11:22 +00:00
jez	538a8969ad	seed the Precedence next generation differently, so can pop as well as just push precedence stack as we unwind imports	2008-11-24 22:26:58 +00:00
jez	6f39b28daa	Reworked current_generation stuff to use the templates' Precedence instead of a plain old int	2008-11-24 22:05:16 +00:00
jez	f610d739fe	More work on rejigging precedence. ... Templates are now constructed with their precedence. Variables are too, with a tweak to allow for the immediate evaluation of non-topl-level params and vars. The execution context now no longer needs to track variable precedence, which is good because it will be getting it wrong anyway. Corresponding simplifications follow to compliation context.	2008-11-19 17:26:07 +00:00
jez	8d0ce36da3	Rework variable handling to use the Precedence class. Currently it has the same (incorrect) behaviour as before, but it's all part of moving to towards (what I hope will be) proper import precedence handling.	2008-11-05 22:55:53 +00:00
jez	e8c0aaa416	Started work on refactoring precedence handling. Need to try and preserver the tree shape it forms, rather than a simple case of monotonically increasing precen ... I mean decreasing precedence.	2008-11-05 02:57:18 +00:00