Since warnings are lexical, you can't control what warnings IO::File will have enabled.
Jeff, when I get a chance I will try it for myself; I don't see when the close of STDIN is actually happening right off.
Does opening STDIN to /dev/null after the open3 call help or not? | [reply] |
The call to open3() opens STDIN for reading in the child process and closes it in the run harness. Adding
# This opens FH0 to prevent File::IOs new 5.8.3 warning
FileHandle->new("</dev/null") or die "unable to open /dev/null: $!"
# Alternatively, open the next file READ as well as write
FileHandle->new("<>$pidFile") or die "unable to open $pidFile: $!"
just after the open3() call prevents File::IO from warning, because FH0 is opened for read.
Ughhh! I will do this for the moment, but this is the first time I can ever recall having to do something unnecessary to suppress a warning. I have, until now always treated warnings as something I NEVER want to do, sigh - IMHO this devalues my 'Absolutely No Warnings' tenet.
Regards,
Jeff
| [reply]
[d/l] |
I don't think it violates your tenet; leaving STDIN closed is really something you never want to do, and that's just what you are avoiding. However, (without having actually looked at it) I don't think the run harness should be closing STDIN when you pass it like that.
I don't think "<>" would work. Try "+<" or "+>".
I would do it as:
open STDIN, "</dev/null" or die ...
so perl's STDIN looks open as well as file descriptor 0.
| [reply]
[d/l] |