Bug #108611 for SNMP-Info: _raw methods not implemented for all subclasses

Mon Nov 09 07:56:33 2015 abraxxa [...] cpan.org - Ticket created

Subject:

_raw methods not implemented for all subclasses

For example in SNMP::Info::Layer3::F5 i_type is overridden but i_type_raw isn't defined. I'd suggest to test all subclasses for interface conformance or use e.g. Role::Tiny to ensure that.

Tue Nov 10 22:08:22 2015 EMILLER [...] cpan.org - Correspondence added

On Mon Nov 09 07:56:33 2015, ABRAXXA wrote: Show quoted text

> For example in SNMP::Info::Layer3::F5 i_type is overridden but > i_type_raw isn't defined. > I'd suggest to test all subclasses for interface conformance or use > e.g. Role::Tiny to ensure that.

A _raw suffix returns data ignoring any munge routines. The is no munge routine for i_type, so i_type_raw will not return anything different than i_type. See http://search.cpan.org/~oliver/SNMP-Info-3.29/Info.pm#AUTOLOAD

Tue Nov 10 22:08:22 2015 The RT System itself - Status changed from 'new' to 'open'

Tue Nov 10 22:08:25 2015 EMILLER [...] cpan.org - Taken

Fri Jan 01 09:27:51 2016 abraxxa [...] cpan.org - Correspondence added

Am Di 10. Nov 2015, 22:08:22, EMILLER schrieb: Show quoted text

> On Mon Nov 09 07:56:33 2015, ABRAXXA wrote:

> > For example in SNMP::Info::Layer3::F5 i_type is overridden but > > i_type_raw isn't defined. > > I'd suggest to test all subclasses for interface conformance or use > > e.g. Role::Tiny to ensure that.

> > A _raw suffix returns data ignoring any munge routines. The is no > munge routine for i_type, so i_type_raw will not return anything > different than i_type. > > See http://search.cpan.org/~oliver/SNMP-Info-3.29/Info.pm#AUTOLOAD

The problem we have with that is that it returns different values than other classes (strings vs. numbers) and so breaks our usage of it. I'd suggest that model specific classes always override the _raw methods and the regular ones just return the munged values as syntactic sugar.

Sat May 14 15:43:36 2016 OLIVER [...] cpan.org - Correspondence added

Hi, Sorry, I don't understand what you're suggesting. The _raw suffix is basically a hint to return data without munging. i_type is an enumerated string, so the raw data will be an integer, and the munged data will be the string types which map to them, from the MIB. All implementations of i_type in SNMP::Info ought to work like this. Or have I missed the point? Perhaps you can give a concrete example? regards, oliver. On Fri Jan 01 09:27:51 2016, ABRAXXA wrote: Show quoted text

> Am Di 10. Nov 2015, 22:08:22, EMILLER schrieb:

> > On Mon Nov 09 07:56:33 2015, ABRAXXA wrote:

> > > For example in SNMP::Info::Layer3::F5 i_type is overridden but > > > i_type_raw isn't defined. > > > I'd suggest to test all subclasses for interface conformance or use > > > e.g. Role::Tiny to ensure that.

> > > > A _raw suffix returns data ignoring any munge routines. The is no > > munge routine for i_type, so i_type_raw will not return anything > > different than i_type. > > > > See http://search.cpan.org/~oliver/SNMP-Info-3.29/Info.pm#AUTOLOAD

> > > The problem we have with that is that it returns different values than > other classes (strings vs. numbers) and so breaks our usage of it. > I'd suggest that model specific classes always override the _raw > methods and the regular ones just return the munged values as > syntactic sugar.

-- regards, oliver.

Sat May 14 15:43:38 2016 OLIVER [...] cpan.org - Status changed from 'open' to 'resolved'

Sun May 15 08:13:50 2016 abraxxa [...] cpan.org - Correspondence added

Hi Oliver, yes, like I initially wrote SNMP::Info::Layer3::F5 defines sub i_type but not sub i_type_raw which leads to i_type_raw not returning the raw values.

Sun May 15 08:14:14 2016 abraxxa [...] cpan.org - Status changed from 'resolved' to 'open'

Wed May 25 08:24:43 2016 OLIVER [...] cpan.org - Correspondence added

Hello again, OK I think I see the misunderstanding here. Because we load the MIB for F5, net-snmp will automatically translate enum values from their integer to the string representation (such as "ethernetCsmacd"). Additionally, SNMP::Info can be implemented with a munge routine to further manipulate the values, but that's not done in this case. The _raw suffix to any method is a hint to avoid this second munge. There is no similar way to avoid the MIB translation of the enum. So is that what you're seeing? If not, please can you give some examples of how the output differs from what you expect to see, as I cannot reproduce it here. regards, oliver. On Sun May 15 08:13:50 2016, ABRAXXA wrote: Show quoted text

> Hi Oliver, > yes, like I initially wrote SNMP::Info::Layer3::F5 defines sub i_type > but not sub i_type_raw which leads to i_type_raw not returning the raw > values.

-- regards, oliver.

Sat Feb 11 05:25:05 2017 OLIVER [...] cpan.org - Correspondence added

Closed and moved to Github: https://github.com/netdisco/snmp-info/issues/211 -- regards, oliver.

Sat Feb 11 05:25:06 2017 OLIVER [...] cpan.org - Status changed from 'open' to 'resolved'