Re: shell variables and attributes
On Thu, May 26, 2022, at 10:39 PM, Christoph Anton Mitterer via austin-group-l at The Open Group wrote: > 2) The draft 2.1 says on page 2398 line 77766: > >The unset utility shall unset each variable or function definition >specified by name that does not >have the readonly attribute and remove any attributes other than >readonly that have been given to >name (see Section 2.14 export and readonly). > > To me that behaviour would seem quite odd ... and it doesn't match the > behaviour of at least bash and dash. > > > What the text currently says is: > `unset` unsets each var/function that is not readonly... > AND (with no further restriction) > removes all attributes other then readonly > (that is: whether or not the var is readonly). > > So that means: >export var=foo >readonly var >unset -v var > as per description, should leave var set (with value "foo"), should > leave the readonly attribute... but should remove the export attribute. Seems that ksh93u+, mksh R59c, and yash 2.52 don't follow this interpretation either. > 3) It's IMO not fully clear whether a readonly variable is allowed to > have other attributes added (or cleared - see (2)). > > On page 2376, line 76926 says: >The values of variables with the readonly attribute cannot be changed >by subsequent assignment or use of the export, getopts, readonly, or >read utilities, nor can those variables be unset by the unset utility. > > and in fact, it seems to work with bash/dash to first mark a var > readonly (whether it's set or not) and then mark it as export. > > But still, given that the other way round seems to not work... it could > IMO deserve further clarification... and maybe emphasis, too, as > allowing to add further attributes is actually a bit contradicting the > read-only property, as it may change the behaviour. At a minimum, bash 5.1.16, ksh93u+, mksh R59c, dash 0.5.11.5, and yash 2.52 already allow exporting an existing readonly variable. If the standard ought to clarify anything, it's that the results of this action are unspecified. -- vq
shell variables and attributes
Hey. I noted the following... 1) First, my understanding is, that a variable needs not to be set, but can still have attributes (e.g. export/readonly) set for it. Not sure whether I just didn't find it, but I think this behaviour is not specifically mentioned somewhere, or is it? 2) The draft 2.1 says on page 2398 line 77766: The unset utility shall unset each variable or function definition specified by name that does not have the readonly attribute and remove any attributes other than readonly that have been given to name (see Section 2.14 export and readonly). To me that behaviour would seem quite odd ... and it doesn't match the behaviour of at least bash and dash. What the text currently says is: `unset` unsets each var/function that is not readonly... AND (with no further restriction) removes all attributes other then readonly (that is: whether or not the var is readonly). So that means: export var=foo readonly var unset -v var as per description, should leave var set (with value "foo"), should leave the readonly attribute... but should remove the export attribute. 3) It's IMO not fully clear whether a readonly variable is allowed to have other attributes added (or cleared - see (2)). On page 2376, line 76926 says: The values of variables with the readonly attribute cannot be changed by subsequent assignment or use of the export, getopts, readonly, or read utilities, nor can those variables be unset by the unset utility. and in fact, it seems to work with bash/dash to first mark a var readonly (whether it's set or not) and then mark it as export. But still, given that the other way round seems to not work... it could IMO deserve further clarification... and maybe emphasis, too, as allowing to add further attributes is actually a bit contradicting the read-only property, as it may change the behaviour. 4) It may not be fully clear, whether unsetting a non-set variable is intended to just clear it's attributes: Line 77766 says: The unset utility shall unset each variable or function definition specified by name that does not have the readonly attribute and remove any attributes other than readonly that have been given to name (see Section 2.14 export and readonly). This could mean that... but OTOH, line 4 says: Unsetting a variable or function that was not previously set shall not be considered an error and does not cause the shell to abort. which rather makes it sound as if this would *just* not be considered an error... (but otherwise have no effect - while in fact it clears the flags). Cheers, Chris.
