Hi,
yikes.

cts is the jdbc compliance test suite from sun. I found it once but lost
it. I don't know how hard it is to find anything in it.

Since I'm working on a ic replacement I'm reluctant to devote much time to
ic.

One possibility for you if you don't want to get too involved in ic
internals might be to write a wrapper driver that wraps ic, replacing the
resultset ( and maybe resultsetmetadata) with classes that map column names
the way you want. JBoss (www.jboss.org) has a generic wrapper to convert
type 1 jdbc drivers to fake XADatasources, although this doesn't replace
resultsets, just the connections.

As a totally different approach, can you just define a lot of views and
select from them? if you have calculated columns you're not going to be
writing to them anyway.

Thanks
david jencks

On 2001.05.23 11:49:21 -0400 Dietrich Schulten wrote:
> Hi David,
>
> David Jencks <davidjencks@...> schrieb:
>
> I'm in a typical go-between situation: OpenOffice's jdbc connection
> (www.openoffice.org) can't work with Interbase calculated columns and
> aliases, and that is the primary "itch" I want to scratch.
> Now the question is: fix the driver or fix the application. Since we're
> talking OSS I'm in this happy situation :-)
>
> In OpenOffice (and BTW StarOffice 5.2, too) the gui database components
> are based on column names. This is astonishing first, but from the
> component user's perspective it makes sense. You want to drop an edit
> field, assign a column name and go on, you don't want to use ordinal
> numbers or draw distinctions between labels and column names. If you have
>
> ambiguous names you use aliases to be sure which field you're referring
> to.
>
> In JDBC the situation seems to be the exact opposite:
>
> "When a getXXX method is called with a column name and several columns
> have the same name, the value of the first matching column will be
> returned. "
>
> -> Never rely on unique names in applications. I've heard this and
> listened
> as well ;-)
>
> But the resultset javadoc continues:
> "The column name option is designed to be used when column names are used
>
> in the SQL query that generated the result set. For columns that are NOT
> explicitly named in the query, it is best to use column numbers. If
> column names are used, there is no way for the programmer to guarantee
> that they actually refer to the intended columns."
>
> -> The question is what that means. It could mean: you should completely
> avoid column names and only use column numbers. It could also mean: while
>
> the primary way are ordinal numbers, you *might* use getXXX(colname) if
> the
> user explicitly named every column in the query using AS.
>
> The JDBC driver mmMySql, for instance, seems to allow getXXX(colname) and
>
> findColumn(colname) for aliased columns. The effect is, there are no
> problems with mySql and Star/OpenOffice as long as you use aliases for
> ambiguous column names.
> In Interclient both colname based methods fail for every aliased column,
> not only calculated columns. IOW, no Interbase calculated columns and no
> aliases in StarOffice.
>
> If you want to offer column names to the end user, you must map ordinal
> numbers, real column names and labels somewhere - either in the driver or
>
> in every application that uses JDBC. The mmMySQL approach to invent the
> wheel only once isn't completely silly, if we set the murky JDBC spec
> aside for a moment.
>
> The key to this is the possibility to use aliases in getXXX(colname) and
> findColumn(colname). The Interclient approach is clean and well, as long
> as you don't want to use a calculated column in a gui component.
>
> At least this whole issue is worth thinking about. A minimal approach
> could be to allow findColumn(colname) and getXXX(colname) to find
> aliases. If I'm not mistaken that would mean first search aliases
> (getColumnLabel), then real column names (getColumnName). E.g. in your
> example where you swapped two columns this would lead to the expected
> result.
>
> > The only other resource to consult I can think of is the cts. Have you
> > looked at it? Pointers to tests there would be appreciated.
>
> I'd like to take a look. What is the cts?
>
> Greetings,
> Dietrich
>
> To unsubscribe from this group, send an email to:
> IB-Java-unsubscribe@egroups.com
>
>
>
> Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
>
>
>
>