Is this a regression for access="stream",form="unformatted"?

[Jerry DeLisle]

FX Coudert wrote:
>> Is it really necessary to revert the patch? Would not it be
>> enough that I open a PR and wait until someone complains
>> for a problem on a real code?
> 
> I agree with Jerry and think it's better to keep a PR open for that 
> enhancement, with a link to the previous patch, and carefully look into 
> it, than having a regression.
> 
> FX
> 
I also have some information from Tobias that may prompt me to some other 
changes in STREAM I/O and I can roll it all together at once.

So I will reopen 33985 and we can add your discovery here to that PR.

Jerry