$ mkdir setpref
[07:18 PM][dagit@apricot~/tmp]
$ cd setpref/
[07:18 PM][dagit@apricot~/tmp/setpref]
$ darcs init
[07:18 PM][dagit@apricot~/tmp/setpref]
$ darcs setpref test "false"
Changing value of test from '' to 'false'
[07:18 PM][dagit@apricot~/tmp/setpref]
$ darcs record -m "setpref"
changepref test

false
Shall I record this change? (1/1)  [ynW...], or ? for more options: y
Finished recording patch 'setpref'
[07:19 PM][dagit@apricot~/tmp/setpref]
$ touch foo
[07:19 PM][dagit@apricot~/tmp/setpref]
$ darcs add foo
d[07:19 PM][dagit@apricot~/tmp/setpref]
$ darcs record -m "foo"
addfile ./foo
Shall I record this change? (1/1)  [ynW...], or ? for more options: y
Finished recording patch 'foo'
[07:19 PM][dagit@apricot~/tmp/setpref]
$ cat ~/.darcs/defaults 
send edit-description
ALL ignore-unrelated-repos

Shouldn't the record of foo fail?

Is this not just a consequence of the default having been switched to --
no-test?

I imagine darcs record --test would have the behaviour you want.

Tentatively marking invalid but do shout if I'm misread something!

I notice that this didn't show up in our release notes, for which, 
sorry.  http://wiki.darcs.net/DefaultSwitches
It may have appeared in pre-release announcements though.  Also, 
changing behaviour does indeed have a cost of confusing previous users.  
It's pain we're voluntarily paying for a shinier future, something we 
can afford a bit more being a minority.  Must not fall into the Git trap 
of cant-change-it-because-people-will-complain.