Fixing regression caused by the 2014-02-19 change ("Fixed long-running

issue ...").

* match.c (open_data_source): if c->data is t, but c->files
is nil, set c->data to nil: we cannot possibly open anything later.
(match_files): We need to call open_data_source one more time just
before processing a line with horizontal material.  The previous
call(s) to open_data_source might not have opened anything. Before
accesing car(c.data) the correct test is consp(c.data), not c.data. In
the else clause, we now specificially check for nilp(c.data) which is
the correct indicator of no more data. If c.data is any other atom at
that point, we have an internal error, for which an assertion is added
now.

1 files changed, 16 insertions, 0 deletions

diff --git a/ChangeLog b/ChangeLog
index 112fffd0..c78565f9 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,5 +1,21 @@
 2014-03-05  Kaz Kylheku  <kaz@kylheku.com>
 
+	Fixing regression caused by the 2014-02-19 change ("Fixed long-running
+	issue ...").
+
+	* match.c (open_data_source): if c->data is t, but c->files
+	is nil, set c->data to nil: we cannot possibly open anything later.
+	(match_files): We need to call open_data_source one more time just
+	before processing a line with horizontal material.  The previous
+	call(s) to open_data_source might not have opened anything. Before
+	accesing car(c.data) the correct test is consp(c.data), not c.data. In
+	the else clause, we now specificially check for nilp(c.data) which is
+	the correct indicator of no more data. If c.data is any other atom at
+	that point, we have an internal error, for which an assertion is added
+	now.
+
+2014-03-05  Kaz Kylheku  <kaz@kylheku.com>
+
 	* stream.c (vformat): Fix broken parsing of parameteric width and
 	precision (i.e. given by *). The simplest way to do this is to add a
 	state vf_star which is similar to vf_digits, and reuses much of the