develooper Front page | perl.perl5.porters | Postings from March 2013

Re: [perl #117265] [PATCH] e213661 no warnings 'safesyscalls', fatal nul checks

Thread Previous | Thread Next
Tom Christiansen
March 23, 2013 22:33
Re: [perl #117265] [PATCH] e213661 no warnings 'safesyscalls', fatal nul checks
Message ID:
>> Second, in the presence of 'use warnings', it instead causes the open to
>> croak (captureable with eval as usual).
>> Is this a correct assessment?

>Yes, the die part is correct.

>> Before discussing the detailed implementation on the patch, I think we
>> need a rough concensus on what semantics we desire.
>> My own personal opinion is that I like the first part: causing system
>> calls to mandatorally (and unchangeably) return failure in the presence of
>> \0.
>> I'm not very keen on the second part. For a start, it doesn't add that
>> much over the first part: the file won't have been opened under any
>> circumstances and writes to it will still fail, even if the return code
>> isn't checked. It's just that suddenly the whole program dies, just because
>> the code contains 'use warnings'.  This seems a big escalation, If people
>> want that behaviour, they can always use autodie.

>There are four possible reactions: ignore, silently strip, warn and stri<SNIP>

>Without warnings or with no warnings 'syscalls' perl behaves as before: <SNIP>

>With warnings enabled I went for die, because such a string argument is 
>almost always a breakin attempt.

Hold on.  You cannot have a fatal warning without asking for one.  

A use warnings pragma must never raise an exception simply because you 
asked for a warning.  That's what "use warnings FATAL => blah" is for.


Thread Previous | Thread Next Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at | Group listing | About