Re: [lmi] Input-sequence editor testing

[Vaclav Slavik]

On Fri, 2010-07-09 at 18:57 +0200, Vaclav Slavik wrote:
> The patch below fixes this 

...and uncovers a bug when inserting a new row that isn't the last one.
This fixes the assert: 

diff --git a/input_sequence_entry.cpp b/input_sequence_entry.cpp
index db209a8..0d8020d 100755
--- a/input_sequence_entry.cpp
+++ b/input_sequence_entry.cpp
@@ -113,6 +113,19 @@ void DurationModeChoice::allow_maturity(bool allow)
         }
     else
         {
+        if(e_maturity == value())
+            {
+            // "until maturity" is selected, but it's no longer allowed.
+            // Replace it with arbitrary other value temporarily; we pick
+            // e_retirement, because it (like e_maturity) doesn't have numeric
+            // argument.
+            //
+            // This is done only to preserve the invariant that
+            // DurationModeChoice always has a value selected; if this happens,
+            // it will be replaced by a more appropriate choice by other UI
+            // code (see UponAddRow).
+            value(e_retirement);
+            }
         Delete(duration_mode_choices - 1);
         }
 }


Regards,
Vaclav