[bind10-dev] death tests (was Re: core dumps on macmini)

[Jelte Jansen]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 12/11/2012 10:02 AM, Michal 'vorner' Vaner wrote:
> Hello
> 
> On Tue, Dec 11, 2012 at 08:54:59AM +0000, Francis Dupont wrote:
>>>> I suspect at least some of them were due to death tests (so
>>>> these are indeed expected results).
>>> 
>>> Do we have some trick to turn them off for death tests? Is it
>>> possible it is forgotten in that places?
>> 
>> => I asked (and opened a ticket) for a way to disable death
>> tests (my issue was with WIN32/VS builds where Release/Debug
>> configs require a different death test setting). So I suggest to
>> apply my proposed change (i.e., apply, review and merge the
>> ticket 2077).
> 
> But we don't want to turn off the death tests, they are there for a
> reason. We want to turn off the core dumps for the death tests.
> 
> (This is, of course, not against the possibility to turn them off
> by a switch).
> 

Unless I misunderstood, the suggestion is just that :) (a switch that
creates a define in config.h that makes death tests be skipped)

There is a whole set of inherent problems with death tests; core dumps
being 'just' one of them. For valgrind checks they also need to be
disabled (though it currently checks at runtime and skips them if
valgrind.h is present on the system, when I added that we also
discussed having some flag or maybe even an environment variable to
skip them).

I'm also in favor of a --disable-death-tests (or possible even the
other way around, disable them by default and add a
- --enable-death-tests switch).

Jelte
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.12 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAlDG+2sACgkQ4nZCKsdOncVEigCgonJYCr4cmpFEpXI/hW/5XYl5
I58An0bYB2/yyX7K9ppTfeJABwrBDoln
=rsnz
-----END PGP SIGNATURE-----