ADBA Add type safe newFactory method

[Lance Andersen]

Consistency with other APIs,

For maximum portability you want to return/encourage use of DataSourceFactory while allowing someone to access MyDataSourceFactory

> On Jul 19, 2018, at 11:26 AM, Douglas Surber <douglas.surber at oracle.com> wrote:
> 
> What is the argument for the String version?
> 
>> On Jul 19, 2018, at 8:18 AM, Lance Andersen <lance.andersen at oracle.com <mailto:lance.andersen at oracle.com>> wrote:
>> 
>> We definitely want  the string based version.
>> 
>> having both factory methods is also OK.  including  an unWrap method is also an option
>> 
>>> On Jul 19, 2018, at 11:07 AM, Douglas Surber <douglas.surber at oracle.com <mailto:douglas.surber at oracle.com>> wrote:
>>> 
>>> I don’t have a strong argument one way or the other. Both String and Class seem fine. The most recent push takes a String but I have no objection to changing it if there is a good reason. I’m in no position to judge how significant an "elegant Kotlin reified generic type extension function” would be. 
>>> 
>>> Douglas
>>> 
>>>> On Jul 18, 2018, at 2:57 PM, Frédéric Montariol <frederic.montariol at gmail.com <mailto:frederic.montariol at gmail.com>> wrote:
>>>> 
>>>> I really like the alternative newFactory static method taking a Class parameter :
>>>> public static <T extends DataSourceFactory> T newFactory(Class<T> clazz) 
>>>> It would allow elegant Kotlin reified generic type extension function !
>>>> 
>>>> But there is something I don't like, this is the fact that DataSourceFactory.newFactory returns null if no corresponding class is found, causing NPE on next method call : factory.builder()
>>>> I would prefer if it throws an Exception.
>>>> 
>>>> Le ven. 13 juil. 2018 à 19:48, Douglas Surber <douglas.surber at oracle.com <mailto:douglas.surber at oracle.com> <mailto:douglas.surber at oracle.com <mailto:douglas.surber at oracle.com>>> a écrit :
>>>> The internal consensus is to change the spec exactly as you suggest and require the explicit cast if that’s what the user wants. I still would like to hear what others think.
>>>> 
>>>>  public static DataSourceFactory newFactory(String name) { … }
>>>> 
>>>>  DataSourceFactory factory = DataSourceFactory.newFactory(“my.adba.MyFactory”);
>>>> 
>>>> or 
>>>> 
>>>>  MyFactory factory = (MyFactory) DataSourceFactory.newFactory(“my.adba.MyFactory”);
>>>> 
>>>> Douglas
>>>> 
>>>> 
>>>> 
>>>>> On Jul 13, 2018, at 9:46 AM, Douglas Surber <douglas.surber at oracle.com <mailto:douglas.surber at oracle.com> <mailto:douglas.surber at oracle.com <mailto:douglas.surber at oracle.com>>> wrote:
>>>>> 
>>>>> This is a good question. I defined it the way I did specifically to avoid the explicit cast. I have no strong opinion on which is better. I’ll ask for opinions internally and I hope other members of this list give their thoughts.
>>>>> 
>>>>> Douglas
>>>>> 
>>>>>> On Jul 13, 2018, at 8:26 AM, Mark Rotteveel <mark at lawinegevaar.nl <mailto:mark at lawinegevaar.nl> <mailto:mark at lawinegevaar.nl <mailto:mark at lawinegevaar.nl>>> wrote:
>>>>>> 
>>>>>> The existing DataSourceFactory.newFactory(String name) has an unchecked cast using a (possibly inferred) type parameter.
>>>>>> 
>>>>>> Would it make sense to add an alternative with the signature:
>>>>>> 
>>>>>> public static <T extends DataSourceFactory> T newFactory(Class<T> clazz)
>>>>>> 
>>>>>> When you use this, you are tied to an ADBA implementation at compile time, but it will make this more type-safe.
>>>>>> 
>>>>>> And maybe the unchecked cast (and type parameter) should be removed from the variant taking a name only.
>>>>>> 
>>>>>> That is, change the method to
>>>>>> 
>>>>>> public static DataSourceFactory newFactory(String name) {
>>>>>>  if (name == null) throw new IllegalArgumentException("DataSourceFactory name is null");
>>>>>>  return ServiceLoader
>>>>>>          .load(DataSourceFactory.class)
>>>>>>          .stream()
>>>>>>          .filter(p -> p.type().getName().equals(name))
>>>>>>          .findFirst()
>>>>>>          .map(Provider::get)
>>>>>>          .orElse(null);
>>>>>> }
>>>>>> 
>>>>>> Any casts would then be the explicit responsibility of the developer, instead of accidental due to type inference.
>>>>>> 
>>>>>> Mark
>>>>>> -- 
>>>>>> Mark Rotteveel
>>>>> 
>>>> 
>>> 
>> 
>> <oracle_sig_logo.gif> <http://oracle.com/us/design/oracle-email-sig-198324.gif>
>>  <http://oracle.com/us/design/oracle-email-sig-198324.gif> <http://oracle.com/us/design/oracle-email-sig-198324.gif>
>>  <http://oracle.com/us/design/oracle-email-sig-198324.gif>Lance Andersen| Principal Member of Technical Staff | +1.781.442.2037
>> Oracle Java Engineering 
>> 1 Network Drive 
>> Burlington, MA 01803
>> Lance.Andersen at oracle.com <mailto:Lance.Andersen at oracle.com>
 <http://oracle.com/us/design/oracle-email-sig-198324.gif>
 <http://oracle.com/us/design/oracle-email-sig-198324.gif> <http://oracle.com/us/design/oracle-email-sig-198324.gif>
 <http://oracle.com/us/design/oracle-email-sig-198324.gif>Lance Andersen| Principal Member of Technical Staff | +1.781.442.2037
Oracle Java Engineering 
1 Network Drive 
Burlington, MA 01803
Lance.Andersen at oracle.com <mailto:Lance.Andersen at oracle.com>