Benefit: there would be a single
preferences file format shared among all platforms.

Cost:
* currently the only platform that uses xml for
preferences is OSX, and Pd relies on its proprietary,
non-portable tool in order to read and save this file.
So it would be necessary to write an XML parser for
reading and writing in Pd, or bring in an XML library
as a dependency.
* can't leverage any of Pd's current parsers or
use them as a guide.  For example, the Pd
message parser splits tokens on spaces-- the
XML preferences file below doesn't fit that
pattern-- the pd parser would interpret that first
line as a single symbol atom.
* the linux preferences file has long-standing code
in Pd for reading/writing preference files.  That would 
be thrown out to create a newer one that
is more difficult to implement.
* without nested tags we lose most of
the power that XML would have provided.  (For example,
nested tags to make the values in "fontmetrics" self
documenting.)

What are the practical problems with using the current
format for the Pd preferences file under Linux?

-Jonathan

<rate>44100</rate>


<flags>-font-weight normal</flags>
<fontmetrics>
    5 11
    6 13
    7 16
    10 19
    14 29
    22 44
</fontmetrics>

> Mensaje telepatico asistido por maquinas.