While tracing down an issue, I tried building druntime tests without release. They fail, on all platforms I tested (Windows, Linux, OS X).

The reason they don't during normal test runs is that the assertion failures are in contracts and invariants, which are disabled when using -release.

I suggest immediately removing the flag for unittest builds (or running them both with release and without), so that we can then work on determining whether the uncovered failures are real bugs or not with auto-tester support.

David

On Mon, 13 Feb 2012 23:20:18 +0100, David Nadlinger <code at klickverbot.at> wrote:

> While tracing down an issue, I tried building druntime tests without release. They fail, on all platforms I tested (Windows, Linux, OS X).
>
> The reason they don't during normal test runs is that the assertion failures are in contracts and invariants, which are disabled when using -release.
>
> I suggest immediately removing the flag for unittest builds (or running them both with release and without), so that we can then work on determining whether the uncovered failures are real bugs or not with auto-tester support.
>
> David
> _______________________________________________
> dmd-beta mailing list
> dmd-beta at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/dmd-beta

Let's not do that now. It is broken since a very long time, there are
missing
import, spurious print statements and recursive invariants. I really want
to
see this getting cleaned up but it's so much that we need to it
systematically.

On 2/13/12 11:33 PM, Martin Nowak wrote:
> Let's not do that now. It is broken since a very long time, there are
> missing
> import, spurious print statements and recursive invariants. I really
> want to
> see this getting cleaned up but it's so much that we need to it
> systematically.

On second thought, you might be right ? I totally misjudged how long the code resp. build system has been in this state.

David