Bug in #! processing - One More Time

[Maxim Sobolev]

Garance A Drosihn wrote:
> At 12:33 PM +0200 2/24/05, Maxim Sobolev wrote:
> 
>> Garance A Drosihn wrote:
>>
>>>
>>> Or we could do #1 for now, and plan to do #4 after 5.4-release.
>>> Or do #1 now in 5.3, and go with some incompatible change (#2
>>> or #3) only in 6.x-current.
>>>
>>> What do people think?  ...
>>
>>
>> I would vote for making #3 and respective /bin/sh changes and
>> MFCing them into 5.4.
> 
> 
> Do you mean "in time for 5.4-release", or
> do you mean "into 5.4-stable, after 5.4-release is done"?

In time for 5.4 if possible, since putting hype aside, 5.4 is going to 
be the first 5.x release which can be seriously recommended for a 
production deployment. Therefore, whatever 5.4 sets is likely to be 
de-facto standard in FreeBSD word during the next 1-2 years.

> (either way would be fine with me, but I would like to see your
> change to kern/imgact_shell.c MFC'ed before 5.4-release, even if
> we don't get #3 done and tested in time for that)

I'll consider that.

>> We don't have that many shell scripts that
>> rely on the previus functionality - ones that in the base system
>> (if any) can be easily fixed, while ones in /usr/ports can be
>> conditionalized on OSVERSION. Removing yet another superfluous
>> difference between FreeBSD and other systems out there is good
>> thing especially considering that BSD-way creates serious
>> problems that can't be resolved without changing semantics anyway.
> 
> 
> The more I think about it, the more I think this is the right idea.

-Maxim