So considering your answer it seems that all is fine, but it isnt. So can
you show me example how will you do it(local on windows). I mean
decoding,encoding,charset type for two messages 1st String + two ints , 2nd
String + long. And you will be sure that it working 100%. Im running out of
ideas...

2012/1/24 Emmanuel Lécharny <elecha...@apache.org>

> On 1/24/12 5:42 PM, Robert Tomczyk wrote:
>
>> So this how i see it.
>> Every step i make in decoder i mean:
>> if(in.prefixedData(x)) // one step = decode next "element"
>> i have to put this data in my DecoderState for specific kind of data ?
>> i.e.
>> MessageAstate,MessageXstate etc.
>> and if next step will return false. Then decoder will be executed again
>> and
>> as in example i will check my specific decoder what kind of data should i
>> "load" (in.getSomething()). Do i get it right ?
>>
>
> Looking at your code, it seems that the doDecode() method is correct.
>
> Basically, what you do for each data structure in your message is :
>
>    if (I have N bytes in the stream where N is the number of bytes I need
> for the data structure I will read)
>      then
>        continue with the next bytes
>      else
>        return false // will wait until enough data is coming
>
> until you have completed the decoding. That's fine, because you always
> check that you will be able to read an enum (1 byte), an integer (4 bytes)
> etc.
>
> Also note that you will start over from the beginning until you have
> enough bytes in the IoBuffer (the IoBuffer will be expanded to receive the
> extra bytes until the doDecode method returns true)
>
> For Strings, you should be careful though : the String <-> bytes
> conversion has to take care of the encoding. On one computer, if you just
> do something like myString.getBytes(), you will get the bytes using a local
> encoding (ISO-8859-1 on Windows, but most certainly UTF-16 on linux, or
> something else, depending on your local env). The best is to decode using
> one specific encoding.
>
>
>
>
> --
> Regards,
> Cordialement,
> Emmanuel Lécharny
> www.iktek.com
>
>

Reply via email to