Megaco woes with ServiceChangeReason = ReasonToken EQUAL VALUE

Micael Karlberg micael.karlberg@REDACTED
Thu Feb 6 09:36:22 CET 2003


Yes, this was the change we had in mind.

/BMK

Peter-Henry Mander writes:
 > Hi Micael,
 > 
 > I have a "solution" which I believe fits the standard wrt. 
 > serviceChangeReason. I've modified:
 > 
 > otp_src_R8B-2/lib/megaco/src/text/megaco_text_gen.hrl
 > 
 > with a change to enc_serviceChangeReason/2 thus:
 > 
 > enc_serviceChangeReason({reason, Val}, State) ->
 >      case Val of
 > 	asn1_NOVALUE ->
 > 	    [];
 > 	[List] when list(List) ->
 > 	    [
 > 	     ?ReasonToken,
 > 	     ?EQUAL,
 > 	     %%enc_Value(List, State) %%%%%%% replace this ...
 >               enc_QUOTED_STRING(List,State) %% ... with this.
 > 	    ]
 >      end.
 > 
 > Now the Service Change always uses a quoted string for 
 > serviceChangeReason, as described in the IETF ABNF file comment.
 > 
 > Pete.
 > 
 > 
 > Peter-Henry Mander wrote:
 > > Hi Micael,
 > > 
 > > Yes, this is what I expect too. Allow me to elaborate. Below is the 
 > > function I
 > > use to issue the service change, with a small modification to match the 
 > > Reason
 > > field in your example:
 > > 
 > > service_change(ConnHandle, Method, Reason) ->
 > >    megaco:call(
 > >       ConnHandle,
 > >       [#'ActionRequest'{
 > >          contextId = ?megaco_null_context_id,
 > >          commandRequests = [
 > >             #'CommandRequest'{
 > >                command = {
 > >                   serviceChangeReq,
 > >                   #'ServiceChangeRequest'{
 > >                      terminationID = [?megaco_root_termination_id],
 > >                      serviceChangeParms = #'ServiceChangeParm'{
 > >                         serviceChangeMethod = Method,
 > >                         serviceChangeReason = ["901 mg cold boot"], %%% 
 > > mod here
 > >                         serviceChangeVersion = 2
 > >                      }}}}]}],
 > >       []
 > >    ).
 > > 
 > > This produces the following message. The serviceChangeReason field 
 > > format has
 > > quotes, similar to your example:
 > > 
 > > MEGACO/1 [10.1.0.200]:2944
 > > Transaction = 1 {
 > >     Context = - {
 > >         ServiceChange = root {
 > >             Services {
 > >                 Method = Restart,
 > >                 Version = 2,
 > >                 Reason = "901 mg cold boot"
 > >             }
 > >         }
 > >     }
 > > }
 > > 
 > > But if I modify the serviceChangeReason to be a safeChar string, containing
 > > digits only, thus:
 > > 
 > > service_change(ConnHandle, Method, Reason) ->
 > >    megaco:call(
 > >       ConnHandle,
 > >       [#'ActionRequest'{
 > >          contextId = ?megaco_null_context_id,
 > >          commandRequests = [
 > >             #'CommandRequest'{
 > >                command = {
 > >                   serviceChangeReq,
 > >                   #'ServiceChangeRequest'{
 > >                      terminationID = [?megaco_root_termination_id],
 > >                      serviceChangeParms = #'ServiceChangeParm'{
 > >                         serviceChangeMethod = Method,
 > >                         serviceChangeReason = ["901"], %% using safeChar
 > >                         serviceChangeVersion = 2
 > >                      }}}}]}],
 > >       []
 > >    ).
 > > 
 > > I get the following result where 901 appears without quotes:
 > > 
 > > MEGACO/1 [10.1.0.200]:2944
 > > Transaction = 1 {
 > >     Context = - {
 > >         ServiceChange = root {
 > >             Services {
 > >                 Method = Restart,
 > >                 Version = 2,
 > >                 Reason = 901
 > >             }
 > >         }
 > >     }
 > > }
 > > 
 > > So I conjecture that if you change line 32 from ["901 mg col boot"] to 
 > > ["901"]
 > > you too will have a similar result. I think that the comment in the ABNF
 > > suggests that even a safeChar string should be formatted with quotes.
 > > 
 > > <rant importance=low>
 > > I humbly believe that the comment about serviceChangeReason should have 
 > > been rewritten as a rule in the IETF's ABNF file to avoid ambiguity, and 
 > > that Note 1 in Annex B.2 has allowed the IETF to be less than rigourous 
 > > with their own ABNF code.
 > > </rant>
 > > 
 > > Pete.
 > > 
 > > P.s. I modified line 9 in your listing 'cos Mozilla didn't like copying the
 > > binary string in ip4Address.
 > > 
 > > Micael Karlberg wrote:
 > >    1 Hi,
 > >    2
 > >    3 I would like to see your erlang message, because when I try,
 > >    4 it works just fine. Here's an example:
 > >    5
 > >    6 Erlang megaco (service change) message
 > >    7 {'MegacoMessage',asn1_NOVALUE,
 > >    8      {'Message',1,
 > >    9           {ip4Address,{'IP4Address',"********",asn1_NOVALUE}},
 > >   10           {transactions,
 > >   11               [{transactionRequest,
 > >   12                    {'TransactionRequest',
 > >   13                        9998,
 > >   14                        [{'ActionRequest',
 > >   15                             0,
 > >   16                             asn1_NOVALUE,
 > >   17                             asn1_NOVALUE,
 > >   18                             [{'CommandRequest',
 > >   19                                  {serviceChangeReq,
 > >   20                                      {'ServiceChangeRequest',
 > >   21                                          [{megaco_term_id,
 > >   22                                               false,
 > >   23                                               ["root"]}],
 > >   24                                          {'ServiceChangeParm',
 > >   25                                              restart,
 > >   26                                              {portNumber,
 > >   27                                                  55555},
 > >   28                                              asn1_NOVALUE,
 > >   29                                              {'ServiceChangeProfile',
 > >   30                                                  "resgw",
 > >   31                                                  1},
 > >   32                                              ["901 mg col boot"],
 > >   33                                              asn1_NOVALUE,
 > >   34                                              asn1_NOVALUE,
 > >   35                                              asn1_NOVALUE,
 > >   36                                              asn1_NOVALUE}}},
 > >   37                                  asn1_NOVALUE,
 > >   38                                  asn1_NOVALUE}]}]}}]}}}
 > >   39
 > >   40 And this is what it looks like pretty-encoded:
 > >   41
 > >   42 MEGACO/1 [124.124.124.222]
 > >   43 Transaction = 9998 {
 > >   44     Context = - {
 > >   45         ServiceChange = root {
 > >   46             Services {
 > >   47                 Method = Restart,
 > >   48                 ServiceChangeAddress = 55555,
 > >   49                 Profile = resgw/1,
 > >   50                 Reason = "901 mg col boot"
 > >   51             }
 > >   52         }
 > >   53     }
 > >   54 }
 > >   55
 > >   56 And this is what it looks like compact-encoded:
 > >   57
 > >   58 !/1 [124.124.124.222]
 > >   59 T=9998{C=-{SC=root{SV{MT=RS,AD=55555,PF=resgw/1,RE="901 mg col 
 > > boot"}}}}
 > >   60
 > >   61
 > >   62 Regards,
 > >   63     /BMK
 > >   64
 > >   65 Peter-Henry Mander writes:
 > >   66 > Good evening Erlang Megaco gurus,
 > >   67 >
 > >   68 > I have a question concerning the interpretation of the
 > >   69 > serviceChangeReason ABNF description (versions one and two *), which
 > >   70 > quotes (**):
 > >   71 >
 > >   72 >    ; A serviceChangeReason consists of a numeric reason code
 > >   73 >    ; and an optional text description.
 > >   74 >    ; A serviceChangeReason MUST be encoded using the quotedString
 > >   75 >    ; form of VALUE.
 > >   76 >    ; The quotedString SHALL contain a decimal reason code,
 > >   77 >    ; optionally followed by a single space character and a
 > >   78 >    ; textual description string.
 > >   79 >
 > >   80 >    serviceChangeReason  = ReasonToken  EQUAL VALUE
 > >   81 >
 > >   82 > ... and a little further down we have (***):
 > >   83 >
 > >   84 >    VALUE                = quotedString / 1*(SafeChar)
 > >   85 >
 > >   86 > The Erlang Megaco implementation follows the VALUE spec, not the
 > >   87 > quotedString requirement as described in the comment. i.e.
 > >   88 >
 > >   89 > MEGACO/1 [10.1.0.200]:2944
 > >   90 > Transaction = 1 {
 > >   91 >     Context = - {
 > >   92 >         ServiceChange = root {
 > >   93 >             Services {
 > >   94 >                 Method = Restart,
 > >   95 >                 Version = 2,
 > >   96 >                 Reason = 901   <-------  1*(SafeChar)
 > >   97 >             }
 > >   98 >         }
 > >   99 >     }
 > > 100 > }
 > > 101 >
 > > 102 > -OR-
 > > 103 >
 > > 104 > MEGACO/1 [10.1.0.200]:2944
 > > 105 > Transaction = 1 {
 > > 106 >     Context = - {
 > > 107 >         ServiceChange = root {
 > > 108 >             Services {
 > > 109 >                 Method = Restart,
 > > 110 >                 Version = 2,
 > > 111 >                 Reason = "901 Cold Boot"  <------- quotedString
 > > 112 >             }
 > > 113 >         }
 > > 114 >     }
 > > 115 > }
 > > 116 >
 > > 117 > But if the comments are followed to the letter (and I understood
 > > 118 > correctly) the Reason should be quoted every time, thus:
 > > 119 >
 > > 120 > MEGACO/1 [10.1.0.200]:2944
 > > 121 > Transaction = 1 {
 > > 122 >     Context = - {
 > > 123 >         ServiceChange = root {
 > > 124 >             Services {
 > > 125 >                 Method = Restart,
 > > 126 >                 Version = 2,
 > > 127 >                 Reason = "901"  <------- quotedString even when
 > > 128 > 1*(SafeChar) matches.
 > > 129 >             }
 > > 130 >         }
 > > 131 >     }
 > > 132 > }
 > > 133 >
 > > 134 > Erlang Megaco is technically correct if the ABNF is stripped of 
 > > comments
 > > 135 > (which is what happens when building the Erlang Megaco stack from the
 > > 136 > spec, the ABNF/ASN.1 compiler won't read comments, or am I
 > > 137 > underestimating Erlang? :-). It looks as if those knowledgable 
 > > fellows
 > > 138 > of the IETF should have specified:
 > > 139 >
 > > 140 > serviceChangeReason = ReasonToken  EQUAL quotedString (instead of 
 > > VALUE)
 > > 141 >
 > > 142 > Am I correct? How do I force Erlang Megaco to _always_ send
 > > 143 > quotedString, not just when there are unSafeChars in the string? 
 > > Is it
 > > 144 > strictly necessary to enforce quotedString?
 > > 145 >
 > > 146 > Help!
 > > 147 >
 > > 148 > Pete.
 > > 149 >
 > > 150 > * Versions 1.03 as in draft-ietf-megaco-3015corr-03.txt and 2.03 
 > > as in
 > > 151 > draft-ietf-megaco-h248v2-03.txt
 > > 152 >
 > > 153 > ** line 6557 in v1.03 and line 7275 in v2.03
 > > 154 >
 > > 155 > *** line 6623 in v1.03 and 7531in v2.03
 > > 156 >
 > > 
 > > 
 > > 
 > > 
 > > 
 > 
 > 

-- 
Micael Karlberg          Ericsson AB, Älvsjö Sweden
Tel:  +46 8 727 5668     EAB/UHK/KD - OTP Product Development
ECN:  851 5668           Mail: micael.karlberg@REDACTED
Fax:  +46 8 727 5775    



More information about the erlang-questions mailing list