I was thinking more along the lines of spring-jdbc for instance. But I
understand what you're talking about.


On 9 June 2014 21:42, Gary Gregory <[email protected]> wrote:

> JDBC is part of the JRE, as such I do not see the _need_ for it to be in a
> separate jar, and personally I certainly do not _want_ it in a new jar.
>
> If you use JDBC, you MUST specify at least one other jar for your driver
> (some driver are delivered as more than on jar, further complicating
> things). Adding another jar on top of that for this or that connection
> pooling is no big deal.
>
> If I had the time, I'd create a -all jar with a much as possible in one
> place (like for our apps at work) that do all logging through Log4j.
>
> In general, for the third party components our products at work deliver
> that are sliced and diced in a billion jars, it's such a pain to tell our
> customer: oh, you want to use this bell or whistle, well, make sure you
> have foo-feature1.jar in on your classpath, a different feature? ah, that's
> in foo-feature2.jar. And gosh forbid that we forget to deliver one of those
> little jar... some customers are not allowed to just fetch some jar from
> the web and slap it on their servers, our product passes some review and
> that's what they are allowed to use, no more. It is also not uncommon for
> POM files and manifests to be incorrect, sometimes jars are not available
> through Maven repos for our builds, and blah blah, I'm done ranting.
>
> Summary: the fewer jars, the better. And no we do not care too much about
> OSGi as we are our own container.
>
> At least in Java 6 I can now add somedir/* to a classpath to pick up all
> jars in a folder, which at least simplifies our apps start/stop scripts.
>
> Gary
>
>
> On Mon, Jun 9, 2014 at 8:49 PM, Matt Sicker <[email protected]> wrote:
>
>> Using DriverManager on its own is rather worthless in this context, and
>> we can't do much without connection pooling. Of course, the stuff that
>> works via a DataSource would work well in theory since that should always
>> be backed by a connection pooling library like dbcp2. However, it might be
>> beneficial to provide support for that in a separate JAR with extra
>> dependencies. That, or make it optional as usual.
>>
>> --
>> Matt Sicker <[email protected]>
>>
>
>
>
> --
> E-Mail: [email protected] | [email protected]
> Java Persistence with Hibernate, Second Edition
> <http://www.manning.com/bauer3/>
> JUnit in Action, Second Edition <http://www.manning.com/tahchiev/>
> Spring Batch in Action <http://www.manning.com/templier/>
> Blog: http://garygregory.wordpress.com
> Home: http://garygregory.com/
> Tweet! http://twitter.com/GaryGregory
>



-- 
Matt Sicker <[email protected]>

Reply via email to