CVS log for xmlschema/XSV/compile/FSM.py

Request diff between arbitrary revisions

merge one fix from main branch

another list optimisation consequence missed

don't work harder than necessary in nd case

Handle base machine nondeterminism via pseudo-parallelism in
subsumption checking

only reset counters when necessary during subsumption check

fix counter bug in subsumption check (maybe)

Note on states that they have an at-risk (of ND screwup) transition
Don't give warning for ND anymore

changed how winning is handled,
in particular when derived max is None

simplify bigloop check a bit,
eliminate particlesHb011.xsd bug by requiring conflicting derived loop
to be outermost

substantial rework of big loop detection, including change
to handling of IncrEdge in derived machine in all circumstances. . .

mark 2 loop check seems OK

mark 2 big loop detection w/o debugging

debugging big loop detection mark 2

remove big loop check on incr case (can't see that it ever would win)
add inverse big loop check

try detecting base infinite loops

use xsvNS for dumping FSMs

accommodate change to final handling

Remove idiosyncratic max check on Incr edges,
move general max check to TestEdge

include lax in edc check,
check conflicts for subsumption

first cut at runtime EDC

simple impl of weakened wildcards

weird wildcard subsumption case

fix UPA bug wrt as it were a+ by slight rearrangement to empty arcs,
test fixed value constraints OK as part of checking edge subsumption

forestall post-error crash

better, i hope, handling of displaying non-ascii qnames

Allow for more than 26 counters in a schema

give up on MaxEdge, handle max checking more like min checking,
implement all groups properly

fix over-counting cases

first attempt at proper handling of numeric exponents during subsumption checking

Having moved the greedy counter/backtracking/resetting code
to a branch, this attempts to re-implement the other changes from
that branch, i.e.
 * short explanation wrt counters
 * distinguish weak and strong overhead loops
 * <all> bug fix
 * Fix kleene + bug (was == *)

hack Log

short explanation wrt counters

distinguish weak and strong overhead loops

<all> bug fix

Implement new complete counter algorithm
 Fix kleene + bug (was == *)

remove some performance bottlenecks

changed min check to plain guard

plus-edge approach to numeric exponents working?

cul-de-sac wrt counters

working on counters

preparing for counters

work on final/block a bit

handle display of subsumption failures locally

protect against missing decl

use unicode properly for names/dumping

package FSM construction and checking and move to FSM

XSV as package: components