Re: config.status[12]: export: SHELL##: is not an identifier

Paul Eggert <[email protected]> Tue, 6 Aug 2024 12:57:09 -0700
Newsgroups gmane.comp.sysutils.autoconf.general
Organization UCLA Computer Science Department
Message-ID <[email protected]>
On 2024-08-06 06:07, Zack Weinberg wrote:
> On Mon, Aug 5, 2024, at 10:32 PM, Paul Eggert wrote:
>> On 2024-08-05 15:24, Bob Friesenhahn wrote:
>>> This one is old and not a "supported" release. It is "omnios-
>>> r151030-6484b6d7b5".
>>
>> If the bug is limited to unsupported releases let's not worry about it.
> 
> This bug was in Illumos core for many years so I don't think we can safely
> assume it isn't in any currently supported OS just because the one person
> who's taken the trouble to tell us about it is on an old machine.

Quite true. Even for OmniOS, we don't know whether the bug is in the 
currently supported releases. If it is present, then it's valid to worry 
about the bug, perhaps by more strongly preferring bash to ksh (since 
there seems to be no practical way to detect the ksh bug).


> It would really
> help if you could test at least one such derivative's current version
> and report whether the bug is present and the value of ${.sh.version}.

+1. When doing so, please use the exact same code you used originally, 
since the bug likely depends on the exact offset of the "ACEOF" string 
from the start of the shell script.