Fastbufs
========

A *fastbuf* is a stream (or file) abstraction optimized for both speed
and flexibility.

Fastbufs can represent many different kinds of objects: regular files, network
sockets, file descriptors in general, or various memory buffers. These objects
are handled by different fastbuf *back-ends.*

Once you have a fastbuf, you can access it by fuctions similar to those of
`stdio.h`, or you can use a variety of fastbuf *front-ends* providing various
formatted operations.

Please keep in mind that fastbufs do not allow arbitrary mixing of reads and
writes on the same stream. If you need to mix them, you have to call @bflush()
inbetween and remember that the file position reported by @btell() points after
the flushed buffer, which is not necessarily the same as after the data you've
really read.

Most fastbuf back-ends also participate in the libucw resource management system.
If you have a resource pool active, newly created fastbufs are automatically tied
to resources in the pool, so when the pool gets cleaned up, the fastbufs are
freed, too. The bclose() function is still available and it removes the tie
as needed.

.Back-ends:
- <<fbparam,Files (parametrized)>>
- <<fbfile,Regular files>>
- <<fbtemp,Temporary files>>
- <<fblim,File fragments>>
- <<fbmem,In-memory streams>>
- <<fbbuf,Buffers>>
- <<fbgrow,Growing buffers>>
- <<fbpool,Memory pools>>
- <<fbatomic,Atomic files>>

.Front-ends:
- <<ffbasic,Basic functions>>

.Other reading:
- <<internal,Internal structure>>
- <<bconfig,Configuring streams>>

ucw/fastbuf.h
-------------

!!ucw/fastbuf.h

ucw/fb-socket.h
---------------

Fastbufs on network sockets with timeouts.

!!ucw/fb-socket.h

ucw/ff-unicode.h
----------------

Reading and writing of unicode characters.

Invalid codes are replaced by `UNI_REPLACEMENT` when reading.

!!ucw/ff-unicode.h

ucw/ff-binary.h
---------------

!!ucw/ff-binary.h