Use Advanced Search to search the entire archive.
[jsr363-experts] Re: PI
- From: Werner Keil <
>
- To: "
" <
>
- Subject: [jsr363-experts] Re: PI
- Date: Wed, 17 Dec 2014 18:59:51 +0100
Btw. looking at the CDI spec in one of CDI JSRs regular IRC chats, version
1.1 of CDI introduced one of the few classes in javax.enterprise.inject.spi
called CDI.
If we introduced any accessor class like SI into the API, then it probably
suited the javax.measure.spi package here, too.
SPI is already the only package with a secondary dependency ("format"
beside the root package) so having another dependency on pretty much every
interface in "quantity" would not be so tragic. Unlike CDI our SPI is
optional.
Werner
On Sun, Dec 14, 2014 at 1:49 PM, Werner Keil
<
>
wrote:
>
>
As mentioned, we could have a few days for "Spec cosmetic" but not only is
>
the actual code tagged as of this weekend (and I would rather defer changes
>
to the code to future stages than try to mess with that or publish new
>
JARs) but PMO naturally has a slight delay before they can really publish
>
the EDR we provide to them.
>
>
If we send it later than Mon or Tue nothing will happen before Christmas.
>
And most likely it may not even show up before the EC F2F. If we do a 30,
>
40 or 60 day review, that we are reasonably free to do. Not every user is a
>
field expert (like many in this EG) so I can't say, what "non-scientists"
>
both in the EC and wider community really care about in the spec. The API
>
is relatively small compared to even JSR 354 not to mention other bloated
>
ones like 310 (with almost 200 classes, just for time, even including all
>
the quantities we don't exceed 60;-D) so I thought 30 days should be enough.
>
>
Having it out when the EC meets would be good to know what we talk about
>
there;-)
>
>
Werner
>
>
On Sun, Dec 14, 2014 at 1:33 PM, Martin Desruisseaux <
>
>
>
wrote:
>
>
>
> Le 14/12/14 21:28, Werner Keil a écrit :
>
> > We could try that for a further stage.
>
> I agree with that. So let just keep in mind that an
>
> implementation-independent SI class may be possible, but I agree that
>
> trying to insert it now may be too risky.
>
>
>
> Martin
>
>
>
>