Re: [pysnmp-dev] Can't get ConstraintsUnion to fail

[Graham H. <gr...@ze...>]

Ilya,

Thanks for the quick work !

I've tried a variety of SYNTAX rules (e.g. "0|2..10", "2..11|13..20",
"0|2..20") and they all now work as I expect.

Cheers,

Graham.

P. S. Now using PySNMP 4.3.2 ...




On 13 August 2016 at 18:24, Ilya Etingof <il...@gl...> wrote:

>
> Hi Graham,
>
> Thank you for your report! That seems to be a bug in pyasn1 which is just
> fixed in master (e.g. pyasn1 0.1.10):
>
>     https://github.com/etingof/pyasn1
>
> Could you please try it out? The other thing is that I’d highly recommend
> using the latest pysnmp release.
>
> On 11 Aug 2016, at 10:07, Graham Hudspith <gr...@ze...> wrote:
>
> Hi
>
> I'm playing with a MIB where I want to experiment with the range of values
> allowed for a particular param.
>
> This is with PySNMP 4.2.5, PyASN1 0.1.9 on Ubuntu 16.04.1 with Python 2.7.
>
> The current syntax for the param is:
>
> SYNTAX Integer32(2..10)
>
> and, indeed, if the agent sends a value outside that range (say, 12), an
> exception is thrown. Great.
>
> The problem is that if I alter that syntax to:
>
> SYNTAX Integer32(0|2..10)
>
> and re-run the bulkCmd(), the operation now succeeds even though the agent
> is still sending an illegal value (12).
>
> When I examine the output from compiling the MIB file, I can see that the
> old code:
>
> someParam = MibTableColumn((1, 2, 3, 4), Integer32().subtype(subtypeSpec=ValueRangeConstraint(2,
> 10))).setMaxAccess("readonly")
>
> has become what seems sensible:
>
> someParam = MibTableColumn((1, 2, 3, 4), Integer32().subtype(
> subtypeSpec=ConstraintsUnion(ValueRangeConstraint(0,0),
> ValueRangeConstraint(2,10),))).setMaxAccess("readonly")
>
> However, when I breakpoint through the code, I can see that it gets into:
>
> pyasn1/type/constraint.py(192)_testValue():
> -> v(value, idx)
> (Pdb) print self
> ConstraintsUnion(ConstraintsUnion(ValueRangeConstraint(0, 0),
> ValueRangeConstraint(2, 10)), ConstraintsIntersection(ConstraintsIntersection(),
> ValueRangeConstraint(-2147483648, 2147483647)))
>
> and this looks wrong to me. I can see that the "inner" ConstraintsUnion
> will fail, but won't the ConstraintsIntersection succeed, meaning that the
> "outer" ConstraintsUnion will also succeed ?
>
> My apologies if this is too terse a description, I can provide more detail
> if you require.
>
> Regards,
>
> Graham
>
>
> --
>
>


-- 


*Graham Hudspith      Zeetta Networks Limited*Software Engineer    Engine
Shed, Station Approach,
gr...@ze...    Temple Meads,
                     Bristol,
                     BS1 6QH
                     United Kingdom
                     zeetta.com <http://www.zeetta.com/>