[Developers] New compilation scripts - todo

[Arni Magnusson]

Recent correspondence regarding -bounds, see below.

Arni



On Fri, 28 Jun 2013, Dave Fournier wrote:

Sure. I don't think it is a big deal. I think it was just to tell tpl2cpp 
to put the ad_exit() code into the source.

Dave



On Thu, 27 Jun 2013, Arni Magnusson wrote:

> I see.
>
> Does that mean you recommend keeping bound_flag hardwired to 1, and 
> maybe we shouldn't bother passing the -bounds option anymore?
>
> Or the other way? I would be happy if you could make a recommendation, 
> based on your design and experience, and we'll take it from there.
>
> Thanks,
>
> Arni
>
>
>
> On Thu, 27 Jun 2013, Dave Fournier wrote:
>
>> I think is was becuase when I first put it in I didn't want to possibly 
>> break code that worked without it. that was a common strategy for 
>> adding almost anything.  Then when it proved to pose no problem I 
>> removed the exception.
>>
>> Dave
>>
>>
>>
>> On Thu, 27 Jun 2013, Dave Fournier wrote:
>>
>>> Hi Dave,
>>>
>>> I was examining the effect of passing -bounds to tpl2cpp/tpl2rem and 
>>> was somewhat surprised that it has no effect.
>>>
>>> In other words, int bound_flag will always be 1, regardless of what 
>>> the user passes.
>>>
>>> The revision history of the Lex files shows this has been the case 
>>> since ADMB was first imported into the the SVN repository in October 
>>> 2008.
>>>
>>> ---
>>>
>>> My questions are:
>>>
>>> (1) Do you think we should make tpl2cpp/tpl2rem produce different C++ 
>>> code depending on the -bounds user option? Maybe having the bound_flag 
>>> always on doesn't slow down the optimized model.
>>>
>>> (2) Do you think we should stop using the -bounds when building safe 
>>> models? It serves no obvious purpose right now, but maybe it's a good 
>>> thing if we later decide to let -bounds create different C++ code.
>>>
>>> I could just start experimenting, but I would very much appreciate 
>>> your comments on this.
>>>
>>> All the best,
>>>
>>> Arni
>>
>