When I run darcs changes |less on OpenBSD darcs segfaults.

A test case is:

darcs init
echo 1 > 1
darcs add 1
darcs record -am 'a commit' 1
darcs changes | echo # this causes darcs writing to a pipe that's not existing

Another test case is the darcs repo. Important here is that the pager has to
block the pipe, because it filles more than one screen.

A gdb stacktrace looks like this:

gdb> bt
#0  0x1c3c7c42 in deRefStablePtrzh_fast ()
#1  0x00000000 in ?? ()

which is not really helpful...

Christian (not Christoph although you can probably count on me making this
mistake again): is this something that can be caught by a test case?

I can surely make a test case out of it. It turns out that this problem has gone
by now. I have not checked which patch is responsible for it though.

Shall I?

The ideal would be to create a test case, show that it fails on an older version
of darcs (without caring about which patch fixes it) so that we know the test is
actually doing something, and also show that it succeeds, which tells us it
won't happen again.

Thanks!

I think we can just resolve this.