Re: Datasheet internals

[Ben Pfaff]

John Darrington <address@hidden> writes:

> A more serious problem that I've noticed, is that if I load a data
> file, then do FILE NEW and try to enter new data, after a file I get
>
>  src/data/case.c:325: case_data_rw_idx: Assertion `idx <
>  c->case_data->value_cnt' failed.
>
> I'm not sure why that is at the moment.

The proximate cause there is that val_cnt is 0, so that
case_data_rw_idx (&cc, 0) tries to access the first value in a
case with no values in it, causing an assertion.  That can be
fixed by putting "if (val_cnt > 0)" above the memset call or by
calling case_data_all_rw instead of case_data_rw_idx.

But then it'll just assert at the case_data_rw call in the for
loop a few lines farther down.  This is because val_cnt shouldn't
be zero: it should be 1 if you added a new numeric or short
string variable.  But the column isn't getting added to the
datasheet.  Why?  The callback isn't getting fired.  Why?  One
reason is a misspelling, I think: the code isn't sure whether the
callbacks are named with underscores or hyphens.  The following
patch helps a bit:

--- gui.orig/src/ui/gui/psppire-dict.c  2007-06-13 21:45:02.000000000 -0700
+++ gui/src/ui/gui/psppire-dict.c       2007-06-13 21:45:16.000000000 -0700
@@ -110,7 +110,7 @@
   object_class->finalize = psppire_dict_finalize;
 
   signals [VARIABLE_CHANGED] =
-    g_signal_new ("variable_changed",
+    g_signal_new ("variable-changed",
                  G_TYPE_FROM_CLASS (class),
                  G_SIGNAL_RUN_FIRST,
                  0,
@@ -123,7 +123,7 @@
 
 
   signals [VARIABLE_INSERTED] =
-    g_signal_new ("variable_inserted",
+    g_signal_new ("variable-inserted",
                  G_TYPE_FROM_CLASS (class),
                  G_SIGNAL_RUN_FIRST,
                  0,
@@ -135,7 +135,7 @@
 
 
   signals [VARIABLES_DELETED] =
-    g_signal_new ("variables_deleted",
+    g_signal_new ("variables-deleted",
                  G_TYPE_FROM_CLASS (class),
                  G_SIGNAL_RUN_FIRST,
                  0,

But it doesn't solve the whole problem.  One oddity is that
delete_variables_callback doesn't delete any columns from the
datasheet, which it seems that it should, but I don't think
that's the issue here.
-- 
Ben Pfaff 
http://benpfaff.org