The Standard ML Basis Library


The IMPERATIVE_IO signature


Synopsis

signature IMPERATIVE_IO

The IMPERATIVE_IO signature defines the interface of the Imperative I/O layer in the I/O stack. This layer provides buffered I/O using mutable, redirectable streams.


Interface

structure StreamIO : STREAM_IO

type vector = StreamIO.vector
type elem = StreamIO.elem

type instream
type outstream

val input : instream -> vector
val input1 : instream -> elem option
val inputN : instream * int -> vector
val inputAll : instream -> vector
val canInput : instream * int -> int option
val lookahead : instream -> elem option
val closeIn : instream -> unit
val endOfStream : instream -> bool

val output : outstream * vector -> unit
val output1 : outstream * elem -> unit
val flushOut : outstream -> unit
val closeOut : outstream -> unit

val mkInstream : StreamIO.instream -> instream
val getInstream : instream -> StreamIO.instream
val setInstream : instream * StreamIO.instream -> unit

val mkOutstream : StreamIO.outstream -> outstream
val getOutstream : outstream -> StreamIO.outstream
val setOutstream : outstream * StreamIO.outstream -> unit
val getPosOut : outstream -> StreamIO.out_pos
val setPosOut : outstream * StreamIO.out_pos -> unit

Description

structure StreamIO : STREAM_IO
This substructure provides lower-level stream I/O, as defined by the STREAM_IO interface, which is compatible with the instream and outstream types, in the sense that the conversion functions mkInstream, getInstream, mkOutstream, and getOutstream allow the programmer to convert between low-level streams and redirectable streams. Typically, the redirectable streams are implemented in terms of low-level streams. Note that StreamIO.outstream is not a functional stream. The only difference between a StreamIO.outstream and an outstream is that the latter can be redirected.

type vector = StreamIO.vector
type elem = StreamIO.elem
These are the abstract types of stream elements and vectors of elements. For text streams, these are Char.char and String.string, while for binary streams, they correspond to Word8.word and Word8Vector.vector.

type instream
The type of redirectable imperative input streams. Two imperative streams may share an underlying functional stream or reader. Closing one of them effectively closes the underlying functional stream, which will affect subsequent operations on the other.

type outstream
The type of redirectable output streams. Two redirectable streams may share an underlying stream or writer. If this is the case, writing or positioning the file pointer on one of them, or closing it, also affects the other.

input strm
attempts to read from strm, starting from the current input file position. When elements are available, it returns a vector of at least one element. When strm is at end-of-stream or is closed, it returns an empty vector. Otherwise, input blocks until one of these conditions is met, and returns accordingly. It may raise the exception Io.

input1 strm
reads one element from strm. It returns SOME(e) if one element was available; it returns NONE if at end-of-stream. It may block, and may raise the exception Io.

After a call to input1 returning NONE to indicate an end-of-stream, the input stream should be positioned after the end-of-stream.

inputN (strm, n)
reads at most n elements from strm. It returns a vector containing n elements if at least n elements are available before end-of-stream; it returns a shorter (and possibly empty) vector of all elements remaining before end-of-stream otherwise. It may block, and may raise the exception Io. It raises Size if n < 0 or if n is greater than the maxLen value for the vector type.

inputAll strm
returns all elements of strm up to end-of-stream. It may block, and may raise the exception Io. It raises Size if the amount of data exceeds the maxLen of the vector type.

canInput (strm, n)
returns NONE if any attempt at input would block. It returns SOME(k), where 0 <= k <= n, if a call to input would return immediately with at least k characters. Note that k = 0 corresponds to the stream being at end-of-stream.

Some streams may not support this operation, in which case the Io exception will be raised. This function also raises the Io exception if there is an error in the underlying system calls. It raises the Size exception if n < 0.

Implementation note:

It is suggested that implementations of canInput should attempt to return as large a k as possible. For example, if the buffer contains 10 characters and the user calls canInput (f, 15), canInput should call readVecNB(5) to see if an additional 5 characters are available.



lookahead strm
determines whether one element is available on strm before end-of-stream and returns SOME(e) in this case; it returns NONE if at end-of-stream. In the former case, e is not removed from strm but stays available for further input operations. It may block, and may raise the exception Io.

The underlying STREAM_IO stream can be used to easily implement arbitrary lookahead.

closeIn strm
closes the input stream strm, freeing resources of the underlying I/O layers associated with it. Closing an already closed stream will be ignored. Other operations on a closed stream will behave as if the stream is at end-of-stream. The function is implemented in terms of StreamIO.closeIn. It may also raise Io when another error occurs.

endOfStream strm
returns true if strm is at end-of-stream, and false if elements are still available. It may block until one of these conditions is determined, and may raise the exception Io.

When endOfStream returns true on an untruncated stream, this denotes the current situation. After a read from strm to consume the end-of-stream, it is possible that the next call to endOfStream strm may return false, and input operations will deliver new elements. For further information, consult the description of STREAM_IO.endOfStream.

output (strm, vec)
attempts to write the contents of vec to strm, starting from the current output file position. It may block until the underlying layers (and eventually the operating system) can accept all of vec. It may raise the exception Io. In that case, it is unspecified how much of vec was actually written.

output1 (strm, el)
writes exactly one element el to strm. It may block, and may raise the exception Io if an error occurs. In that case, it is unspecified how much of el was actually written, especially if its physical representation is larger than just one byte. At this level, more than this cannot be guaranteed. Programs that need more control over this possibility need to make use of more primitive or OS-specific I/O routines.

flushOut strm
causes any buffers associated with strm to be written out. It is implemented in terms of StreamIO.flushOut. The function may block, and may raise the exception Io when an error occurs.

closeOut strm
flushes any buffers associated with strm, then closes strm, freeing resources of the underlying I/O layers associated with it. It is implemented in terms of StreamIO.closeOut. A write attempt on a closed outstream will cause the exception Io{cause=ClosedStream,...} to be raised. It may also raise Io if another error occurs (e.g., buffers cannot be flushed out).

mkInstream strm
constructs a redirectable input stream from a functional one. The current version of strm returned by input operations will be kept internally and used for the next input. They can be obtained by getInstream.

getInstream strm
returns the current version of the underlying functional input stream of strm. Using getInstream, it is possible to get input directly from the underlying functional stream. After having done so, it may be necessary to reassign the newly obtained functional stream to strm using setInstream; otherwise the previous input will be read again when reading from strm the next time.

setInstream (strm, strm')
assigns a new functional stream strm' to strm. Future input on strm will be read from strm'. This is useful for redirecting input or interleaving input from different streams, e.g., when handling nested include files in a lexer.

mkOutstream strm
constructs a redirectable output stream from a low-level functional one. Output to the imperative stream will be redirected to strm.

getOutstream strm
flushes strm and returns the underlying StreamIO output stream. Using getOutstream, it is possible to write output directly to the underlying stream, or to save it and restore it using setOutstream after strm has been redirected.

setOutstream (strm, strm')
flushes the stream underlying strm, and then assigns a new low-level stream strm' to it. Future output on strm will be redirected to strm'.

getPosOut strm
returns the current position in the stream strm. This raises the exception Io if the stream does not support the operation, among other reasons. See StreamIO.getPosOut.

setPosOut (strm, pos)
sets the current position of the stream strm to be pos. This raises the exception Io if the stream does not support the operation, among other reasons. See StreamIO.setPosOut.

See Also

BinIO, ImperativeIO, STREAM_IO, TextIO

Discussion

A word is in order concerning I/O nomenclature. We refer to the I/O provided by the IMPERATIVE_IO signature as Imperative I/O, while the I/O provided by the STREAM_IO signature is called Stream I/O. On the other hand, the type of buffered I/O handled by both of these layers is typically considered ``stream I/O,'' which explains why the I/O objects defined in both levels are called instream and outstream. To avoid confusion, we sometimes refer to I/O using the Stream I/O layer as functional, focusing on the functional flavor of the input streams at that level. This, however, glosses over the imperative nature of output at the same level. The principal distinction between the two layers is that I/O using IMPERATIVE_IO can be redirected, while I/O using STREAM_IO cannot.

The semantics of Imperative I/O operations are (almost) all defined in terms of the operations provided by the underlying STREAM_IO substructure. Specifically, we have the reference implementations:

fun input(f) = let 
      val (s,g) = StreamIO.input(getInstream f)
      in setInstream(f,g); s
      end
fun inputAll(f)= let 
      val (s,g) = StreamIO.inputAll(getInstream f)
      in setInstream(f,g); s
      end
fun endOfStream(f)= StreamIO.endOfStream(getInstream f)
fun output(f,s) = StreamIO.output(getOutStream f, s)
fun flushOut(f) = StreamIO.flushOut(getOutStream f)
with similar implementations for other imperative I/O operations.

Alternatively, we can consider Imperative I/O streams as ref cells referring to STREAM_IO streams:

type instream = StreamIO.instream ref
type outstream = StreamIO.outstream ref

fun input strm = let 
      val (v, strm') = StreamIO.input(!strm)
      in
        strm := strm'; v
      end
fun output (strm, v) = StreamIO.output(!strm, v)
etc.

The one exception to the above approaches is input1. If an implementation relies solely on StreamIO.input1, input1 could never advance beyond an end-of-stream. To avoid this, a reference implementation for input1 would be:

fun input1 f = let
      val (s,g) = StreamIO.inputN(getInstream f, 1)
      in setInstream(f,g);
         if length s = 0 then NONE
         else SOME(sub(s,0))
      end

Limited random access on input streams --- that is, returning to a previously scanned position --- can be accomplished using getInstream and the underlying Stream I/O layer:

fun reread (f : instream, n : int) = let
      val g = getInstream(f)
      val s = inputN (f,n)
      in
        setInstream(f,g);
        (s, inputN (f,n))
      end
The pair of vectors returned by reread will always be identical. Similarly limited random access on output streams can be done directly using getPosOut and setPosOut. More general random access is only available at the Primitive I/O level.
Implementation note:

Input on a closed stream behaves as though the stream is permanently at end-of-stream. Thus, in addition to closing the underlying functional stream, the closeIn function must also replace the functional stream with an empty stream.


[ Top | Parent | Contents | Index | Root ]

Generated April 12, 2004
Last Modified January 29, 1997
Comments to John Reppy.


This document may be distributed freely over the internet as long as the copyright notice and license terms below are prominently displayed within every machine-readable copy.

Copyright © 2004 AT&T and Lucent Technologies. All rights reserved.

Permission is granted for internet users to make one paper copy for their own personal use. Further hardcopy reproduction is strictly prohibited. Permission to distribute the HTML document electronically on any medium other than the internet must be requested from the copyright holders by contacting the editors. Printed versions of the SML Basis Manual are available from Cambridge University Press. To order, please visit www.cup.org (North America) or www.cup.cam.ac.uk (outside North America).