Re: while loop with extdebug + debug trap always returning 1

[Rocky Bernstein]

Thanks for giving me a heads up here. I have updated the bashdb documentation for the skip command make clear that $? is no changed in skipping an instruction.

More generally bash saves and restores some state on entry and exit to extdebug, like the variable $? (the extdebug function can issue a command which inside sets $? but on return $? I think is restored) 

However there is other global state that isn't saved and restored. I think this includes the set flags. I'm not sure that what gets saved and restored is fully documented, but it probably should be. 

On Fri, Apr 7, 2017 at 9:06 PM, Chet Ramey <chet.ramey@case.edu> wrote:

On 4/4/17 5:08 PM, nesro wrote:

>
> To have both commands similar, let's assume the while as this:
>
> while (( 1 > 2 )); do :; done
>
> Thanks for explaining me why it happen, but now I don't know why for cycle
> works and while does not.
When extdebug is enabled, the command is skipped if the DEBUG trap returns
non-zero.  This doesn't change $? (traps never do) and it doesn't cause the
command that would have been run to return a non-zero exit status.  If it
did, the combination of DEBUG and `set -e' would exit the shell, for
example, and I'm sure it would not behave as the bash debugger would
expect.  For that matter, why should a command that's not run fail?

Since that command returns successfully, when it's used as the `while'
command test, it causes the while command to execute forever.

--
``The lyf so short, the craft so long to lerne.'' - Chaucer
                 ``Ars longa, vita brevis'' - Hippocrates
Chet Ramey, UTech, CWRU    chet@case.edu    http://cnswww.cns.cwru.edu/~chet/