From: Reece Dunn (msclrhd_at_[hidden])
Date: 2003-12-16 15:43:24
Beman Dawes wrote:
>Discussion included ways of providing user customizable formatting for such
>This strikes me as the same problem as the much discussed (on Boost) issue
>of user-customizable serialization formats.
>I wonder what those who have been following Robert Ramey's serialization
>proposal closely think? Could one mechanism solve both serialization and
Serialization can come in several flavours, depending on the storage type
you are writing to:
 XML - how is the data stored in the XML markup?
To support binary serialization, you'll need a specialist serialization
library (like have been posted to Boost and discussed) that has a special
interface (i.e. does not use an I/O stream) in order to gain the maximum
performance and to support binary data.
For the text and XML formats, with respect to use in displaying a sequential
container, surely integration with the I/O streams would provide a more
natural interface for users. Also, in this respect, customizable formatting
would provide the most flexibility for the user, while providing reasonable
defaults for people who - as commented by Paul Bristow - "just want output".
My Output Formatters library has been written to tackle the issue of output
of sequential contaners to a stream. I wonder how my library compares to
Robert Ramey's in this context?
I have added an example to the boost-sandbox to show how my library can be
used for text and XML-based serialization in an output-only context. I have
supplied an experimental implementation for the input of a data type: this
is experimental and incomplete.
I would like some feedback regarding these additions and whether it would be
worthy continuing this line of development.
Reece H Dunn
Sign-up for a FREE BT Broadband connection today!
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk