Bug in datestr.m

[Bill Denney]

datestr.m will give errors if you happen to give a vector or matrix of 
datenums that happens to have 6 columns.  The attached patch fixes this 
and adds a test.

I put the changelog in the changeset.  Is that the correct way to do it now?

Have a good day,

Bill

# HG changeset patch
# User address@hidden
# Date 1202960784 18000
# Node ID 50626ab21494f69e82d811a9ff1552f3b1f1af44
# Parent  8d6ab12f8fda3d17be69c515a726716af0c88ca7
datestr.m: fix 6 datenum vector bug

diff -r 8d6ab12f8fda -r 50626ab21494 scripts/time/datestr.m
--- a/scripts/time/datestr.m    Sat Feb 09 03:37:39 2008 -0500
+++ b/scripts/time/datestr.m    Wed Feb 13 22:46:24 2008 -0500
@@ -203,7 +203,20 @@ function retval = datestr (date, f, p)
   if (iscell (date) || columns (date) != 6)
     v = datevec (date, p);
   else
-    v = date;
+    v = [];
+    if columns (date) == 6
+      ## make sure that the input really is a datevec
+      maxdatevec = [inf 12 31 23 59 60];
+      for i = 1:numel (maxdatevec)
+        if any (date(:,i) > maxdatevec(i))
+          v = datevec (date, p);
+          break;
+        endif
+      endfor
+    endif
+    if isempty (v)
+      v = date;
+    endif
   endif
 
   for i = 1:(rows (v))
@@ -328,6 +341,9 @@ endfunction
 %!assert(datestr(testtime,29),"20051218");
 %!assert(datestr(testtime,30),"20051218T023317");
 %!assert(datestr(testtime,31),"2005-12-18 02:33:17");
+## avoid the bug where someone happens to give a vector of datenums that
+## happens to be 6 wide
+%!assert(datestr(733452.933:733457.933), ["14-Feb-2008 22:23:31";"15-Feb-2008 
22:23:31";"16-Feb-2008 22:23:31";"17-Feb-2008 22:23:31";"18-Feb-2008 
22:23:31";"19-Feb-2008 22:23:31"])
 # demos
 %!demo
 %! datestr (now ())