www-jcp-open mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Craig L Russell <Craig.Russ...@Sun.COM>
Subject Re: JSR 275 - I would like to vote no
Date Mon, 25 Jan 2010 18:29:50 GMT

On Jan 25, 2010, at 6:38 AM, Geir Magnusson Jr. wrote:

>
> On Jan 25, 2010, at 8:58 AM, Andrus Adamchik wrote:
>
>> Hi Geir,
>>
>> I am not currently working with the problem domain touched by this  
>> JSR, although I have some science background (a degree in  
>> chemistry) and understand exactly where the need for this is coming  
>> from.
>
> me too.  I'm also/was a physicist, which is why some of this stuff  
> rubbed me the wrong way (Mass isn't weight, for example...)

As I read it, Mass is in the standard, weight is only used as class,  
variable, and method names in examples (not normative). They have  
several examples of Weight as a subclass of Mass so they can use  
Weight as a type in Person. Most developers (physicists excluded, of  
course) would model a Person's mass as Weight, not Mass.
>
>> One day is not enough time to evaluate its technical merit, but  
>> what I DON'T like is seeing the JRE standard library expand adding  
>> yet another package ("javax.measure") that is too specialized.
>
> Yes
>
>>
>> Could you clarify something in this regard. Say this JSR goes  
>> final. Does it automatically become a part of J2SE, or is it  
>> possible for it to exist as a standalone package, and who defines  
>> each of the outcomes?
>
> It is not automatically a part of J2SE, but it will be *very*  
> difficult to replace once it "owns" the units/measure space, and to  
> me, this is somewhat fundamental.
>
>>
>> I'd be -1 for the former and +1 for the later (unless I overlooked  
>> some other warning signs in the API). I.e. this may (or may not)  
>> turn out to be a fine library for scientists, but let's not put it  
>> in the JRE.
>
> So of the problems, there is no time and date in this, and there  
> should be.

Methinks there is another JSR for time and date. This JSR is already  
complex just dealing with the relatively (compared to time/date)  
issues of reconciling different units of measurement.

> There are oddities like constants in interfaces, things are named  
> strangely, it's been argued that the API doesn't play well with  
> existing APIs, etc.
>
> If you look at the initial vote :
>
> http://www.jcp.org/en/jsr/results?id=3216
>
> You'll see plenty of concerns and a real honest intent that rather  
> than run to standardization, they EG was encouraged to get broad  
> input, and create an implementation and USE IT in running code for  
> real scientific and other workloads to see if it really works....

Was this ever done? If so, where is it? If not, it's a good reason to  
reject it (EGs should take EC input seriously).

Craig
>
> geir
>
>
>
>>
>> Andrus
>>
>> On Jan 25, 2010, at 3:26 PM, Geir Magnusson Jr. wrote:
>>> JSR 275 is a "Unit Specification" JSR.  I've seen two  
>>> presentations on the spec, and I'm not comfortable at a technical  
>>> level that this is on the right track.
>>>
>>> However, I represent the ASF (and not me), so I'd like some  
>>> feedback on the spec - talk me out of it if you want :)
>>>
>>> Sorry about the short timeframe - I have to vote tonight...
>>>
>>> http://www.jcp.org/en/jsr/detail?id=275
>>>
>>>
>>
>

Craig L Russell
Architect, Sun Java Enterprise System http://db.apache.org/jdo
408 276-5638 mailto:Craig.Russell@sun.com
P.S. A good JDO? O, Gasp!


Mime
View raw message