I like the structure of streams wrapping streams from the point of view of
modularity of implementation and flexibility of reuse (as in applying buffering
and encoding to in-memory or network streams).
But I don’t like exposing al the machinery to the user. For example, instead of
bufferedReadStream := ZnBufferedReadStream on: aReadStream.
bufferedWriteStream := ZnBufferedWriteStream on: aWriteStream.
why not
bufferedReadStream := ZnBuffer on: aReadStream.
bufferedWriteStream := ZnBuffer on: aWriteStream
where the class that implements the buffering looks at the stream and decides
whether to do read or write buffering (or both, for a read-write stream).
The meaning of #position: on an encoded Unicode stream is perfectly
well-defined. It’s just not the same as the meaning of position on the
underlying byteStream — one counts in codePoints, the other in bytes. The old
file streams got this wrong; I’m hoping that the new ones get it right.
However, the comment that
"As a side note: using #size and #position[:] on variable encoded
streams
like UTF8 is *very* dangerous and should be avoided.”
makes me worry that they do not. Please, either implement #size and
#position[:] correctly on the encoded stream, or don’t implement them at all!
Do *not* implement them incorrectly, and then tell me to avoid them because
they are dangerous!
Andrew
