Subject: Re: package update.
From: Rob Anderson (rob@reawebtech.com)
Date: Thu Oct 03 2002 - 10:03:38 PDT
David,
I'd appreciate if you could do some regressions like you said,
good idea. What simulators do you have?
I assume Jim and I have Model T.
I still need test cases for the new functions, Jim (hint, hint).
I can combine the test cases and send them to David for posting
Can someone clarify the new IEEE stance about distributing the
package? Paul? Just curious, I heard it could now be distributed
but our page says otherwise.
Thanks,
Rob
David Bishop wrote:
> "Paul J. Menchini" wrote:
>
>>Jim Lewis wrote:
>>
>>
>>>Are you thinking of appending a year code: numeric_std_2002 ?
>>>
>>>I think this is a good idea as it helps with portability of old code.
>>>
>>I would not support a change to the name of the package in the absence
>>of factors that would break old code. Specifically, if none of the
>>following statements are true
>>
>>* features are removed
>>
>>* the semantics of existing features are changed in a non-upward
>>compatible fashion
>>
>>I would not change the name. Otherwise, we'll break old code for no
>>reason. (In the absence of a name change, unless one of the above
>>statements were true, old code would exhibit no changes in behavior.)
>>
>
> That being the case, we should place another constraint on ourselves:
>
> Do not change the functonality of the current packages, and do not add
> functionality which would necessitate legacy code to be changed.
>
> We should run the old test suite aginst the new code when we are done
> to verify this. Note that I have a copy of that verification code sitting
> in a subdirectory of mine just in case.
>
>
This archive was generated by hypermail 2b28 : Thu Oct 03 2002 - 10:02:47 PDT