>>There seems to be a serious problem (bug?) in the behavior of views
>>defined programmatically - they seem to work fine most of the time, but
>>once you get any kind of error, it starts behaving in most unimaginable
>>ways.
>
>It seems to be related to development time problem. In production, once
>that is all debugged, it works ok.
The ODBC connection to Excel is quite unstable. First, it doesn't work
with programmatically defined views (it seems that there's not a way to
check the "include system tables" checkbox programmatically). Second, if
the user makes any change to the structure of the sheet, all is gone to
hell.
Supposing that in all other cases the remote table structures are
sacred, you're right. Though, the ODBC error codes are no better than
the old Fortran error codes - just numbers, no detailed description,
doesn't state the offending argument. Debugging remote views
is close to nightmare, but when they work, they work fine.