Re: [dtn-interest] new i-d on sdnvs

["Gerardo Izquierdo" <[e-mail]>]

Hi Wesley

I've been working a bit more with the code and found why it works even if it
returns the null string.

Problem is that an empty null-terminated string consist in just an 0x00 wich
can be read as an SNDV of value 0.

Perhaps we can have troubles in systems wich encodes the string in a
different way.

In fact, the length of the string encoding a SNDV with it's 7 final bits
equal to 0 will be different in lenguajes using null-terminated strings
(length -1)

That is not a problem itself, but we should be aware in order to avoid
problems when we convert null-terminated strings to non null-terminated
strings.
By example in Borland Pascal lets be Cstring a null-terminated string and
PString a Pascal style string, then if Cstring contains an SNDV you should
code the conversion carefully

Pstring := StrPas(Cstring); {Built-in function to convert between
null-terminated strings and Pascal-style strings}
if ord(Pstring[Length(Pstring)])>127  {if the high-order bit of the last
byte is 1}
   then Pstring := Pstring + chr(0);   {then append a 0x00}

Also, if using null-terminated-strings, when constructing the headers, it's
possible to have problems with an extra byte to 0x00

Please if anyone thinks this subject is too much specific for the list, feel
free in sending me a direct-email.

(And yes, I program mainly in Pascal, and yes, I'm a fossil 8-D )

Gerardo

On 4/24/06, Wesley Eddy <[e-mail]> wrote:
>
> On Mon, Apr 24, 2006 at 05:28:43PM +0200, Gerardo Izquierdo wrote:
> > I think the python Source Code about SNDV is  a bit incorrect
>
> You're right, thanks.  I think maybe the draft should state that the code
> is just for demonstration purposes, and that (for example) the C++ SDNV
> code in the DTN2 implementation should more similar to what people
> actually
> use in production.  That said, it is also important the the code be
> correct
> in this document.
>
> > sdnv_encode does not encode well the value 0 (0x00), it returns an empty
> > string
> >
> > I haven't never programmed in python, but I think we should add a line
> just
> > before the
> > return r
> >
> >
> > if type(n) == type(1) and n = 0:
> >    r = chr (0x00)
> >
>
> Instead of doing this, I changed the looping construct to always execute
> at least once:
>    done = False
>    while not done:
>      # encode lowest 7 bits from n
>      newbits = n & 0x7F
>      n = n>>7
>      r = chr(newbits + flag) + r
>      if flag == 0:
>        flag = 0x80
>      if n == 0:
>        done = True
>
>
> > also, there is no reference to the optional parameter slen in
> sdnv_decode
> > but in the parameter declaration.
>
> The line:
>    elif i == len(s)-1:
> should be:
>    elif i == len(s)-1 or i > slen:
>
> These will be fixed in the next update.  Thanks.
>
> -Wes
>
> --
> Wesley M. Eddy
> Verizon Federal Network Systems
>
>
>