One of my favorite sayings: "If you don't make mistakes, you're not
trying." What's important is to learn, and I think rhere is a general
lesson here. The change to @thin semantics for @file changed the
semantics of the .leo file; i.e., earlier versions of Leo could no
longer read the .leo file. One can therefore not rely on new code to
generate - or even preserve - a test case that it could not naturally
generate itself (I'm sure there's a way to hide a test .leo file from
a containing .leo file and write it out, but a read-only blob sitting
in the test tree would be simpler). I.e., transitions are hard, and
need to be tested.
- Stephen
On Feb 24, 10:26 am, "Edward K. Ream" <[email protected]> wrote:
> The problems with @file that have just resurfaced show how hard it can
> be to manage complex engineering projects. I got blind-sided by this
> problem, despite the fact that it had been dealt with before.
>
> The problem was compounded by the fact that the unit tests for @file
> never failed. Looking at them just now, I see that they silently
> became irrelevant sometime during the course of 4.7 development.
> Indeed, the unit tests generate file-like sentinels, and then verify
> that Leo writes those sentinels as expected. But the new @file ==
> @thin code forces thin-like sentinels. Presto, the old unit tests
> began to test the wrong thing.
>
> This shows, quite clearly, that unit tests can create a false sense of
> security. If there is an automated way to avoid this trap I don't
> know what it is.
>
> Edward
>
> P.S. In order to make progress, I relentlessly focus on closing
> issues. But any mistake in doing so can create a time bomb. One such
> bomb has just exploded.
>
> Here, with Leo, the stakes are relatively small. With the Space
> Shuttle the stakes are much bigger, and the engineering difficulties
> much larger. It is all too easy to criticize the management mistakes
> after they came into sharp focus after the fact. At the time,
> however, the mistake was only one of thousands of decisions. Could
> any of us honestly say we surely would have done better? I doubt it.
>
> EKR
--
You received this message because you are subscribed to the Google Groups
"leo-editor" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to
[email protected].
For more options, visit this group at
http://groups.google.com/group/leo-editor?hl=en.
