inline as well

2014-12-29 17:21 GMT+01:00 Anatole Tresch <[email protected]>:
> Hi Romain/all
>
> see inline
>
> 2014-12-29 16:58 GMT+01:00 Romain Manni-Bucau <[email protected]>:
>
>> Hello Anatole,
>>
>> Tried to put all my thoughts regading current api modules:
>>
>> - Configuration:
>>
>> * looking it I feel like
>> org.apache.tamaya.Configuration#get(java.lang.String) should be the
>> only method of Configuration. I'm not sure about current(), still
>> think it shouldnt be here but in something like in
>> ConfigurationFactory but not in configuration itself - at least to be
>> aligned on common practises
>
>
> -1 for String only, type safe access is one of the most wanted feature in
> the community.
> -1 for putting it elsewhere: It is JDK 8 style to do so. On top it would
> require adding an additional artifact to the API, which is not necessary.
> Keepo the API as small as possible. If seee further use cases that require
> additional artifacts, we may reconsider this, but as of now I would let be
> where it is.
>

excepted it is not consistent this way: why providing some specific
method but not "mine". Why allowing to get implicit conversion and to
provide a PropertyAdapter as parameter.

etc...

I agree it is an important and nice feature but it is not a core one.

One thing you forgot is: as a user you want *your* logic which can be
spring, xbean, a custom one.... That's why i strongly think it should
be outside the core to ease integration with other frameworks and not
collide on conversion process.

About current(): all EE stack does it otherwise ;). The fact JDK 8
allows to mix layers doesnt mean we should do it and that's what does
current for me: mixing service and data layers.

>
> * All "adapt-friendly"  methods should go elsewhere IMO but not the
>> core API - I guess it could be a mecanism used by all libs we'll add
>> on top of it but I wouldn't bind tamaya-core-api to it since today
>> when you integrate a config with an existing factory you already have
>> it and it just adds noise the core ATM - once again I dont say to drop
>> it but just to move it to another module for now.
>>
> -1 see above (one of the most wanted features). Additionally getXXX for
> JDK wrapper types is a quite common API style in SE. Perhaps Werner can
> correct me, if I am wrong here. Definitively it helps users, since they
> now, that for these types the have guaranteed adapter support and this is
> clearly visible from the API.
>

Maybe where we don't agree is the fact I'm sure nobody excepted
framework writers will use this layer so smaller it is better it is. I
would put the easiness effort on next layer and not this one.

>
>> - PropertySource:
>>
>> * not sure I get name need, for a future gui?
>>
> Yes and no, Name allows traceability and enables dynamic access of the
> configuration components from outside, e.g. from managment functionality.
> Name could be the file name and additional info.
>

using toString can be enough while we don't need this feature (why
introducing an API if the framework does nothing with it?).

>
>
>> * think it misses something between get(String) and getProperties() -
>> == I agree with the TODO to replace getProperties() by
>> getPropertyKeys(). Wonder if we should get PropertySource and
>> ListablePropertySource instead of isScannable(). Last one would
>> implement Iterable<String> (for getPropertyKeys()). wdyt?
>>
> I would prefer as well adding the keySet() method (it was part of my
> original proposal).
> -1 for Iterable<String>: I can do that easily from the key set, and I don't
> want to have references on my property source through whatever loops at
> runtime (possible risk of side effects).
> -1 for ListPropertySource. Most of the property sources, even remote ones,
> will be scannable. Using instanceof to check, if an instance is scannable
> or not, seems for me to much efforts, compared to having the small boolean
> method.
>

Was not the point. Point was on implementer side API is easier to use.
Impl is a detail then.

> - PropertySourceProvider:
>>
>> * a detail but shouldn't it be PropertySourcesProvider?
>>
> tbd (other opinions?). In Deltaspike it is called ConfigSourceProvider.
> For me it looks OK, also with the singular...
>
>
>
>> - ServiceContext:
>>
>> * do we need all these default methods? I would do T
>> getTamayaService(Class) and List getUserServices(Class) maybe. This
>> can have an impact on the way it is loaded then - you dont want to
>> load some tamaya services from a lower classloader (SericeContext and
>> Configuratoin[Factory] ones typically). ATM getService() and
>> getServices() doesn't differentiate it and then we could combine both
>> in our usages and have issues later
>> .
>>
>
> We know there are things to improved here. This class was not yet analyzed
> in more detail last night (it was about 1:30 pm !). We will continue
> today...
>
> Anatole
>
>
>
>>
>>
>>
>
>>
>>
>>
>> Romain Manni-Bucau
>> @rmannibucau
>> http://www.tomitribe.com
>> http://rmannibucau.wordpress.com
>> https://github.com/rmannibucau
>>
>>
>> 2014-12-29 16:11 GMT+01:00 Anatole Tresch <[email protected]>:
>> > Hi Oliver/all
>> >
>> > when you look at the api package you will find the API we have agreed on
>> > together last night. Lookt at it and feel free to discuss. We will meet
>> for
>> > another Hangout session this evening again.
>> >
>> > -Anatole
>> >
>> > 2014-12-28 20:21 GMT+01:00 Oliver B. Fischer <[email protected]>:
>> >
>> >> I hate timezones: 9 PM UTZ at Tuesday...
>> >>
>> >> Am 28.12.14 um 19:55 schrieb Werner Keil:
>> >>
>> >>> UTZ, so does it mean Midnight CET?
>> >>> And did "Thuesday" mean Tuesday or Thursday?;-)
>> >>>
>> >>> I can't say, if I'll be on a tablet then. This is a prepaid, but I may
>> top
>> >>> it up although I'm just here till the first week of Jan to use that.
>> >>> Other places I use there are people long after midnight in most cases,
>> but
>> >>> that depends on public transport in Vienna going that long...
>> >>>
>> >>> Everything but Wed night sounds OK, but even Tue and Thu people will
>> >>> already/still launch a lot of fireworks here;-O
>> >>>
>> >>> Werner
>> >>>
>> >>> On Sun, Dec 28, 2014 at 7:23 PM, Oliver B. Fischer <
>> >>> [email protected]
>> >>>
>> >>>> wrote:
>> >>>> What's about Thuesday around 11pm UTZ?
>> >>>>
>> >>>> Am 28.12.14 um 16:42 schrieb Mark Struberg:
>> >>>>
>> >>>> --
>> >>>> N Oliver B. Fischer
>> >>>> A Schönhauser Allee 64, 10437 Berlin, Deutschland/Germany
>> >>>> P +49 30 44793251
>> >>>> M +49 178 7903538
>> >>>> E [email protected]
>> >>>> S oliver.b.fischer
>> >>>> J [email protected]
>> >>>> X http://xing.to/obf
>> >>>>
>> >>>>
>> >>>>
>> >> --
>> >> N Oliver B. Fischer
>> >> A Schönhauser Allee 64, 10437 Berlin, Deutschland/Germany
>> >> P +49 30 44793251
>> >> M +49 178 7903538
>> >> E [email protected]
>> >> S oliver.b.fischer
>> >> J [email protected]
>> >> X http://xing.to/obf
>> >>
>> >>
>> >
>> >
>> > --
>> > *Anatole Tresch*
>> > Java Engineer & Architect, JSR Spec Lead
>> > Glärnischweg 10
>> > CH - 8620 Wetzikon
>> >
>> > *Switzerland, Europe Zurich, GMT+1*
>> > *Twitter:  @atsticks*
>> > *Blogs: **http://javaremarkables.blogspot.ch/
>> > <http://javaremarkables.blogspot.ch/>*
>> >
>> > *Google: atsticksMobile  +41-76 344 62 79*
>>
>
>
>
> --
> *Anatole Tresch*
> Java Engineer & Architect, JSR Spec Lead
> Glärnischweg 10
> CH - 8620 Wetzikon
>
> *Switzerland, Europe Zurich, GMT+1*
> *Twitter:  @atsticks*
> *Blogs: **http://javaremarkables.blogspot.ch/
> <http://javaremarkables.blogspot.ch/>*
>
> *Google: atsticksMobile  +41-76 344 62 79*

Reply via email to