I missed the end of the presentation due to my son hitting me in the head with a tennis racket! Just watched the bit I missed.
One thing I noticed when converting my cfcunit tests to mxUnit was the setup method had to change from private to public
So my setup function begins: <cffunction name="setUp" returntype="void" access="public">
Try that on your unit test and see if it helps your setup() method.
Alan
________________________________________
From: mxu...@googlegroups.com [mxu...@googlegroups.com] On Behalf Of Peter Bell [pb...@systemsforge.com]
Sent: 08 May 2008 22:50
To: mxu...@googlegroups.com
Subject: [mxunit:414] Possible install error
Thanks. I'd have to check out the recording again, but I've got a
feeling we just did something silly like not put cfscript tags around
our code or something, because I just dropped the User =
createobject() into setup now, pulled out the createobject from the
methods, and all the tests passed (I deliberately broke it to get them
all red and then fixed to get them green again to make sure the green
was for real.
The joys of coding live in a presentation!
Thanks for turning up. Hope it made some sense . . .
Best WIshes,
Peter
I'm wondering if the dreaded cfcexplorer being run in the "install"
page is somehow messing with CF, scrambling its brain or something.
this was a bug fixed in the 8.01 updater: 70782 When a mapping was set
for "/", complete component and interface paths
were note resolved correctly.
Peter, any chance you're still running without the updater and have a
"/" mapping?
I wonder if you installed the updater if your need to "hack" the
index.cfm file would go away.
Marc
Glad the mystery is revealed! Thanks re: the pair programming. I think
the approach really works for "remote test infection" and the pairing
with Ben has certainly helped me to get more into the habit of "test
first" - something neither of us were doing consistently before.
Best Wishes,
Peter
Oh well . . .
Best Wishes,
Peter