Re: [Wonder-disc] Oracle date prototype generating to_timestamp sql statement

[Amiel M. <bos...@gm...>]

Now why do you always have the good stuff :)

Luckily there is an er.extensions.ERXJDBCAdaptor.className property to
override JDBCAdaptor class, but nonetheless, there are a number of classes
needed to be subclassed as well to get to the JDBCColumn class. :(

Amiel

On Wed, Oct 13, 2010 at 9:09 AM, Mike Schrag <ms...@po...> wrote:

> boo ... yeah, not in your version .... that sucks ... kind of hard for you
> to do a custom one :\ some hackery with reflection in a JDBCChannel
> subclass, probably, is the only way. or jad all of JDBCColumn and replace it
> with an override.
>
> On Oct 12, 2010, at 5:55 PM, Amiel Montecillo wrote:
>
> Hey Mike,
>
> I'm not seeing a defaultColumnClass in EROraclePlugin or OraclePlugin. Its
> only defaultExpressionClass that I can override.
>
> Thanks,
> Amiel
>
> On Tue, Oct 12, 2010 at 10:17 PM, Mike Schrag <ms...@po...> wrote:
>
>> ah ok .. clears up that craziness.
>>
>> oracle made a mess of dates, but one thing you need to be careful of is
>> that the value type you use matches the column type you declared in oracle,
>> or it won't use an index ... basically it doesn't use an index on any data
>> type conversion.
>>
>> if you're setting value type 'D', though, i would already expect that it's
>> doing a setObject(..., Types.DATE) though. regardless, though, you can
>> override defaultColumnClass on your PlugIn subclass and return a custom
>> column subclass.
>>
>> as an aside, be careful about taking sql with merged bind variables as an
>> accurate representation of your queries. oracle optimizes differently for
>> the bind variable form of a sql statement compared to the non-bind variable
>> version. so if you're testing those sql statements in sql developer, it's
>> generally better to use bind variables there as well.
>>
>> ms
>>
>> On Oct 12, 2010, at 6:52 AM, Amiel Montecillo wrote:
>>
>> Hi Mike,
>>
>> Actually I saw the "to_timestamp" with the use of log4jdbc
>> http://code.google.com/p/log4jdbc/.
>>
>> I felt the need to go beyond EOAdaptorDebugEnabled to print out the sql
>> statement just before being pushed to the DB connection.
>>
>> With log4jdbc, you can see the evaluated sql statement instead of the
>> "withBindings" logging from EOAdaptorDebugEnabled.
>>
>> Thanks,
>> Amiel
>>
>> On Tue, Oct 12, 2010 at 9:45 PM, Mike Schrag <ms...@po...> wrote:
>>
>>> Soooo I don't even see any code that generates "to_timestamp". Any reason
>>> you aren't using bind variables? The bind variables path through
>>> OraclePlugIn already does a statement.setObject(x, timestamp,
>>> date/timestamp) based on valueType.
>>>
>>> ms
>>>
>>> On Oct 12, 2010, at 1:20 AM, Amiel Montecillo wrote:
>>>
>>>
>>> Tracked this down to
>>> com.webobjects.jdbcadaptor.JDBCColumn.takeInputValue(). So it is always a
>>> java.sql.Timestamp that goes to the preparedstatement.
>>>
>>> Is there a way to patch JDBCColumn with my own subclass?
>>>
>>> I'd like to change the Date is handled so that if the valueType is "D",
>>> it will use java.sql.Date instead of Timestamp.
>>>
>>> Any idea is greatly appreciated.
>>>
>>> Thanks,
>>> Amiel
>>>
>>> On Mon, Oct 11, 2010 at 10:54 AM, Amiel Montecillo <bos...@gm...>wrote:
>>>
>>>> Hi,
>>>>
>>>> I am debugging some slowness on my queries with oracle 11g.
>>>>
>>>> I have an "endDate" DATE column in one of my tables.
>>>>
>>>> In my model.plist:
>>>>
>>>> {
>>>>             allowsNull = Y;
>>>>             columnName = endDate;
>>>>             name = endDate;
>>>>             prototypeName = date;
>>>>             userInfo = {"_EntityModeler" = {generateSource = NO; }; };
>>>>             valueType = D;
>>>> },
>>>>
>>>> EOF generates:
>>>>
>>>> SELECT t0.id FROM Bid t0 WHERE t0.endDate < *to_timestamp*('10/11/2010
>>>> 00:00:00.000', 'mm/dd/yyyy hh24:mi:ss.ff3')
>>>>
>>>> I was expected EOF to generate a "to_date" sql statement but instead I
>>>> get the statement above thus performing a full table scan instead of making
>>>> use of the endDate index.
>>>>
>>>> Am I missing something?
>>>>
>>>> Thanks,
>>>> Amiel
>>>> --
>>>> socket error: unable to connect to 127.0.0.1
>>>>
>>>
>>>
>>>
>>> --
>>> socket error: unable to connect to 127.0.0.1
>>>
>>> ------------------------------------------------------------------------------
>>> Beautiful is writing same markup. Internet Explorer 9 supports
>>> standards for HTML5, CSS3, SVG 1.1,  ECMAScript5, and DOM L2 & L3.
>>> Spend less time writing and  rewriting code and more time creating great
>>> experiences on the web. Be a part of the beta today.
>>>
>>> http://p.sf.net/sfu/beautyoftheweb_______________________________________________
>>> Wonder-disc mailing list
>>> Won...@li...
>>> https://lists.sourceforge.net/lists/listinfo/wonder-disc
>>>
>>>
>>>
>>
>>
>> --
>> socket error: unable to connect to 127.0.0.1
>>
>>
>>
>
>
> --
> socket error: unable to connect to 127.0.0.1
>
>
>


-- 
socket error: unable to connect to 127.0.0.1