As I was learning about the darcs test suite today, I was confused by 
the output of test TODOs. Eg:

$ perl merging_newlines.pl
ok 1 - reality check: manifest
ok 2 - reality check: darcs w -s
ok 3 - expect conflicts when pushing
not ok 4 - after a conflict, darcs resolve should report a conflict # 
TODO waiting on code to fix this
#   Failed (TODO) test 'after a conflict, darcs resolve should report a 
conflict'
#   at merging_newlines.pl line 48.
#                   'No conflicts to mark.
# '
#           matches '(?i-xsm:no conflicts)'
1..4

When run en masse, it looks like this:

merging_newlines.......................1/?
#   Failed (TODO) test 'after a conflict, darcs resolve should report a 
conflict'
#   at merging_newlines.pl line 48.
#                   'No conflicts to mark.
# '
#           matches '(?i-xsm:no conflicts)'
merging_newlines.......................ok

So as I understand it the merging_newlines test passed, but printed this 
(scary) output as a reminder to developers. My reaction as a newbie 
darcs tester is that these don't belong here, because

- they make it harder for testers to see whether tests passed

- they nag too many people - testers should not be bothered with random 
things that developers hope to do some day

- they are another mechanism for documenting tests known to fail, which 
we already have in bugs/*

I would propose that these kinds of TODO items be made silent 
(developers know how to grep) and/or be moved to the bugs/* tests. What 
do you think ?

-Simon

_______________________________________________
darcs-users mailing list
[email protected]
http://lists.osuosl.org/mailman/listinfo/darcs-users

Reply via email to