(j3.2006) (SC22WG5.4672) [ukfortran] AW: Informal WG5 ballot on new draft DTS

[Malcolm Cohen]

Bill Long writes:
>I'm inclined to say that someone who tries this deserves to have their program 
>not work.

Obviously we are all agreed on that.

This discussion seems to have gone up several garden paths despite Bill's 
attempt to bring it back on topic.

NOTE 8.11 is about changing the C descriptor itself, not whether you do 
undefined things by computing with the members thereof.  That is a totally 
different topic, and should not be discussed in *THIS* note.

Bill previously wrote:
>My main interest is whether we all agree with the  wording change
viz
>>>>"A C function
>>>> that modifies a C descriptor other than as permitted by this
>>>> Technical Specification will cause undefined behaviour."

Obviously I agree with that.

Further, if Reinhold/Nick/anyone else wants to lecture C programmers on how not 
to abuse their newfound Fortran interoperability, fine, but in a separate note 
please.  Note also that notes are informative not normative, and so therefore 
never "essential"; indeed, they must be supported by normative text.  And in 
almost all of the sidetracking discussion, the violations are violations of the 
base Fortran standard anyway, so not particularly on topic for the TS anyway. 
And the C programmer can already make a complete mess of things by starting with 
TYPE(C_PTR), so much of this goes back to F2003.

Cheers,
-- 
................................Malcolm Cohen, Nihon NAG, Tokyo.