How exactly would users benefit from a half-change? Seems like nothing
but a headache noticing which ones worked and which did not.
Setting multiple items at once implies a transaction IMO.
On Oct 2, 2009, at 3:57 PM, Sreenivas Munnangi wrote:
> In situations like that should we enable the user to express his
> intent thru boolean option --transaction, meaning all the values
> have to be set atomically or not.
>
> Just my idea.
>
> Hans Hrasna wrote:
>> I think 3 is what most people would expect.
>>
>> Personally I think 4 is the most desirable.
>>
>> Option 1 is better than nothing, in my opinion. The user should be
>> warned that both "c" and "e" were not set.
>>
>> Bill Shannon wrote:
>>> I asked a few people privately but I thought it would be good to get
>>> more opinions...
>>>
>>> If I do
>>>
>>> asadmin set a=b c=d e=f
>>>
>>> and "c" is an invalid name, what do you expect to happen?
>>>
>>> 1. fail after setting "a"?
>>>
>>> 2. fail after setting "a" and "e"?
>>>
>>> 3. fail before setting anything?
>>>
>>> 4. succeed with a warning that "c" wasn't set?
>>>
>>>
>>> Right now (with my pending changes) it does #1 because that's what
>>> was easy to implement.
>>>
>>>
>>> ---------------------------------------------------------------------
>>> To unsubscribe, e-mail: dev-unsubscribe_at_glassfish.dev.java.net
>>> For additional commands, e-mail: dev-help_at_glassfish.dev.java.net
>>>
>>>
>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: dev-unsubscribe_at_glassfish.dev.java.net
>> For additional commands, e-mail: dev-help_at_glassfish.dev.java.net
>>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe_at_glassfish.dev.java.net
> For additional commands, e-mail: dev-help_at_glassfish.dev.java.net
>
Lloyd Chambers
lloyd.chambers_at_sun.com
GlassFish Team