> Hi All,
>
[quoted text clipped - 27 lines]
> then I'll read some bytes
> with orginal InputStream?
> Hi All,
>
[quoted text clipped - 12 lines]
> ints - futher more
> it reads int in big-endian order (the transmission is little-endian).
First of all you should always stick to the internet standard of network
byte order, which is big endian. Second of all write the size in text
instead of as bytes, the same goes with the binary data format it using
Base64 coding or something similar. The you can read and process
everything as text and then convert to byte,integer, float etc, what you
need to convert.
Basically, stick to one format or the other, i.e. binary or text. It
doesn't complicate things as much and it will be easier to extend the
design of the protocol.
> I pass socket.getInputStream() into constructor of my reading thread:
>
[quoted text clipped - 11 lines]
> then I'll read some bytes
> with orginal InputStream?
you should never mix streams. It makes the code unmanageable and the
functionality untrustworthy. Unless you are chaining them, as you have
done with the this.in stream. And yes it is technically possible to open
several readers and manipulate them independently, but as said its not
advisable.
An example is using a PushbackReader and when you find any bytes, you
push it back and open a new InputStream from the pushback stream and
read the raw bytes. But then you are going to have problems the next
time you read from the BufferedReader, because the mark of the stream is
at different places depending on whether you are using the inputstream
stream or the BufferedReader stream.
So as stated it is better to choose either complete binary or text protocol.
tom
EJP - 18 Oct 2006 01:40 GMT
> An example is using a PushbackReader and when you find any bytes, you
> push it back and open a new InputStream from the pushback stream
Well you can't even do that for a start, you can't put an InputStream on
top of a Reader.
In the OP's circumstance I would just use the dreaded
DataInputStream.readLine() to read the lines and DataInputStream.read()
to read the bytes.
Tom Forsmo - 18 Oct 2006 01:51 GMT
>> An example is using a PushbackReader and when you find any bytes, you
>> push it back and open a new InputStream from the pushback stream
>
> Well you can't even do that for a start, you can't put an InputStream on
> top of a Reader.
No, but you can use the InputStream the Reader is based. Ok, its not
opening up a new InputStream, but you are still using the underlying
stream directly.
tom
EJP - 20 Oct 2006 04:27 GMT
>>> An example is using a PushbackReader and when you find any bytes, you
>>> push it back and open a new InputStream from the pushback stream
[quoted text clipped - 5 lines]
> opening up a new InputStream, but you are still using the underlying
> stream directly.
... but you still won't get any of the data pushed back into the
PushbackInputStream!
EJP - 20 Oct 2006 04:28 GMT
>>> An example is using a PushbackReader and when you find any bytes, you
>>> push it back and open a new InputStream from the pushback stream
[quoted text clipped - 5 lines]
> opening up a new InputStream, but you are still using the underlying
> stream directly.
... but you still won't get any data pushed back into the PushbackReader!
Tom Forsmo - 21 Oct 2006 02:02 GMT
> .... but you still won't get any data pushed back into the PushbackReader!
I went back to the paragraph in the book I thought I read this, but I
must have misread it. I apologise for the confusion on that claim.
tom
Tom Forsmo - 21 Oct 2006 11:36 GMT
>>>> An example is using a PushbackReader and when you find any bytes,
>>>> you push it back and open a new InputStream from the pushback stream
[quoted text clipped - 7 lines]
>
> .... but you still won't get any data pushed back into the PushbackReader!
I went back to the paragraph in the book I thought I read this, but I
must have misread it. I apologise for the confusion on that claim.
What you can do, though, is to use a PushbackInputStream. Your read some
bytes, give it to a reader which parses it, when the reader reaches the
binary data, it returns with a value stating how many bytes where
consumed by the reader. What was not consumed is pushed back into the
stream and you open up e.g. a DataInputStream to read the binary data.
When the DataInputStream is finished you could potentially return and
give control to the reader again, this way going back and forth between
text and binary. Its a bit complicated and not as clean. It would be
easier and cleaner to just use a complete binary or text based protocol
instead.
tom