Michael,

> Hi all, esp whoever uses the initials "yo" or "jm".

  Oh, well, "yo" is me and "jm" is John Maloney.  (He is also known as
'jhm'...)  But, I think I'm not as guilty as he is for this matter^^;

> There's a bit of (what I consider, as an unqualified newbie) untidyness 
> in PositionableStream. Under the method categories "nonhomogenous 
> access" and "data get/put", there are a whole bunch of rather useful 
> methods. Some of these are duplicated, further proving that they are 
> needed and useful. These are namely things like "uint32" and "uint32:" 
> which retrieve and put unsigned 32-bit integers on a stream.

  Yes, some of them are redundant.

> My argument for these being in the class of the respective data types 
> (SmallInteger etc) is that it is the class that should know about the 
> format of that data structure in bytes, and not a stream.

  Probably.  This is hard to explain, but if you've got the spec of a
message packet from someone that looks like:

  | header <int8> | type <uint16> | nField <uint32> | ...

and want to fill the packet with some values, I would want to write a
method look like:

    fillPacketWithHeader: header type: type nField: nField
        aStream int8: header.
        aStream uint16: type.
        aStream uint32: nField.

because the method seems to concentrate aStream more on those values,
and it is easier to see the result conforms the format.

  You can still say:

    fillPacketWithHeader: header type: type nField: nField
        header writeInt8On: aStream.
        type writeUint16On: aStream.
        nField writeUint32On: aStream.

But it at least looks longer^^;

  I don't know about making them primitives...  Why do you want them
be primitives?

> Furthermore, if all objects supported something like a writeOn: method, 
> serialization would be easy. References are tricky though - they would 
> probably call "aStream writeReferenceOf: self" so that the Stream can do 
> fancy things with references (like ImageSegments :-) ).

  Hmm.  I am kind of ok with the current situation.  DataStream
implements bunch of #writeBlah: so you can take a look at a single
method category and figure out what it does.

-- Yoshiki