From: Ian Jackson <ian@liberator.relativity.greenend.org.uk>
Date: Sun, 12 Dec 2010 01:20:49 +0000 (+0000)
Subject: layout/data2safety: improve TODO.slip-multi-movfeat plan
X-Git-Url: http://www.chiark.greenend.org.uk/ucgi/~ijackson/git?a=commitdiff_plain;h=40704278759e51305fabad21c508dc03a8f99465;p=trains.git

layout/data2safety: improve TODO.slip-multi-movfeat plan
---

diff --git a/TODO.slip-multi-movfeat b/TODO.slip-multi-movfeat
index da09b00..64141b2 100644
--- a/TODO.slip-multi-movfeat
+++ b/TODO.slip-multi-movfeat
@@ -1,8 +1,15 @@
 * new section in ours.wiring for breaking up S0 S1 S2 S3 into
   P0Q0V0R0 et al.
-
 * data2safety reads that and computes most things to have the
   exploded version; the exception is layout-data.c SegPosCombInfo.pname
+	bidirectional mapping constructed from "movfeatposmap" lines
+	map from ([A-Za-z]+[0-9]+)* to ([A-Za-z]+[0-9]+)*
+	has two effects:
+		- in line_endwiring, used "forwards"
+		- when writing out .pname, used "backwards"
+	in each case we match only whole feature positions
+		by some suitable regexp trick
+	we record whether a mapping entry was used, and complain if not
 
 * movpos.c needs to be modified to cope with mixed MovPosChanges
   split MovPosChange into two structs, Change which is per-method