[Openswan dev] Re: PATCH: fix oops during KLIPS unloading

[D. Hugh Redelmeier]

| From: Paul Wouters <paul at xelerance.com>

| On Mon, 13 Feb 2006, D. Hugh Redelmeier wrote:
| 
| > | From: Paul Wouters <paul at xelerance.com>
| >
| > | The bug item is left open until a testcase has been written for this issue.
| >
| > The methodology calls for a test BEFORE a fix.  That is the way you
| > can tell that the test actually will fail without the fix (i.e. that
| > the fix fixes what the test tests).
| >
| > When you get around to writing the test, be sure to revert this fix so
| > that you can test the test.
| 
| The fix did see some testing on Iron. So in that sense it is confirmed.
| I agree a testcase ideally is written before committing code.

Read what I said more carefully: it is the test that will not be
tested properly unless you go out of your way to revert the fix.

Down the road, this will threaten the integrity of the fix.  The fix
may fall out, unnoticed, if the test isn't right.

Tricky stuff.  The guidelines are there for your safety.

MCR has been saying this all along.  I'm just trying to explain why he
has been saying this.  So maybe folks will actually obey :-)