<div class="gmail_quote">On 10 May 2011 18:39, Roger Leigh <span dir="ltr"><<a href="mailto:rleigh@codelibre.net">rleigh@codelibre.net</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
The problem is, from the POV of the keyfile parser, it really *is*<br>
unknown: the chroot object reads all the keys it knows about, and<br>
anything left over is unknown. chroots not supporting unions don't<br>
posess the union "facet" which provides the unioning functionality,<br>
and so they simply are unable to process that key. The chroot<br>
objects are (internally) extensible via the addition of "facets",<br>
each of which can process arbitrary keys; in consequence it's not<br>
possible to know at the time of the error that an "unknown" key is<br>
valid in other contexts.<br></blockquote><div><br></div><div>In that case, maybe just improvements to the man page. i.e.</div><div><br></div><div>1. List which types support unions. Or list which types don't support unions.</div>
<div><br></div><div>2. Also mention that the default type is 'plain'. I see a reference that says 'directory' is recommended but unfortunately it fails to mention that the default is 'plain'.</div>
<div><br></div><div>If the documentation was acceptable on this point, I would have been less confused.</div><div><br></div><div>Also maybe the wording of the message could be improved. e.g.</div><div><br></div><div>Unknown key ‘union-type’ used for type 'plain'</div>
<div><br></div><div>Just to emphasis that the the type may be relevant to the error.</div></div>-- <br>Brian May <<a href="mailto:brian@microcomaustralia.com.au">brian@microcomaustralia.com.au</a>><br>