On Tue, Jul 21, 2009 at 7:17 PM, Frans Pop<elendil@xxxxxxxxx> wrote:
> On Tuesday 21 July 2009, Steve Calfee wrote:
> I enabled all the debug print statements in your script; what I get is:
> 1 /busnum
> 2 /busnum
> 3 /busnum
> 4 /busnum
> 5 /busnum
> 6 /busnum
> 7 /busnum
> 8 /busnum
> 9 /busnum
> 10 /busnum
> [... ad infinitum ...]
>
> So it looks like it's already going wrong in the very first loop in the
> BEGIN function.
>
> My sysfs looks fairly normal, nothing weird connected at all. And there
> are only 7 hubs listed:
> $ ls -d1 /sys/bus/usb/devices/usb*
> /sys/bus/usb/devices/usb1
> /sys/bus/usb/devices/usb2
> /sys/bus/usb/devices/usb3
> /sys/bus/usb/devices/usb4
> /sys/bus/usb/devices/usb5
> /sys/bus/usb/devices/usb6
> /sys/bus/usb/devices/usb7
>
> This is with gawk from Debian stable ("Lenny"):
> $ gawk --version
> GNU Awk 3.1.5
>
Hi Franz - thank you very much for testing this. There was a operator
precedence change between gawk 3.1.5 and my version 3.1.6 between the
implicit concatenation operator and the pipe "|" operator. So I
changed the piped parts to have explicit precedence, downloaded the
old gawk and tested it. So now it should work for you too.
> So it looks as if there's either a bug in 3.1.5, or you're doing something
> that was not supported in that version. But allowing a script to run wild
> in that way is not really good programming anyway I'd say.
I agree. Testing is hard and regression testing is even harder. Thanks
for your feedback.
> BTW, did you test what happens with your script if /sys/bus/usb/devices/
> does not exist or when there are no hubs in it? I made sure that the
> usb-devices script included some sanity checks for those cases.
>
I added an explicit exit in that case, thanks for the suggestion.
Regards, Steve
Attachment:
usbd.awk
Description: Binary data
