• H. S. Teoh (17/17) Feb 27 2013 I just had a sinking realization today: I have been using assert inside
• Andrej Mitrovic (10/14) Feb 27 2013 Are you sure? The asserts work for me:
• Andrej Mitrovic (17/20) Feb 27 2013 Btw I think this might be an accident, in mars.c there's this

"H. S. Teoh" <hsteoh quickfur.ath.cx> writes:

I just had a sinking realization today: I have been using assert inside
unittests to check for test results, but actually, this is wrong!

Why? Because when you compile with -release -unittest, all those asserts
disappear, and the unittests become useless.  The correct way is to use
enforce instead of assert.

Unfortunately, Phobos unittests are full of asserts rather than enforce.
:-/

(And I contend that compiling with both -release and -unittest is a
valid scenario: it lets you catch bugs that only show up in the release
version, such as accidentally relying on side-effects in asserts, which
work when compiling without -release, but mysteriously fail during
-release. Failing unittests would allow you to catch such mistakes
before you ship it to the irate customer.)


T

-- 
Computerese Irregular Verb Conjugation: I have preferences.  You have
biases.  He/She has prejudices. -- Gene Wirchenko

"Andrej Mitrovic" <andrej.mitrovich gmail.com> writes:

On Thursday, 28 February 2013 at 04:52:01 UTC, H. S. Teoh wrote:
 Why? Because when you compile with -release -unittest, all 
 those asserts
 disappear, and the unittests become useless.

Are you sure? The asserts work for me:

unittest
{
     int getZero() { return 0; }
     assert(getZero != 0);
}

void main() { }

$ dmd -release -unittest test.d
$ test.exe
 core.exception.AssertError test(6): unittest failure

"Andrej Mitrovic" <andrej.mitrovich gmail.com> writes:

On Thursday, 28 February 2013 at 04:58:48 UTC, Andrej Mitrovic 
wrote:
 $ dmd -release -unittest test.d
 $ test.exe
 core.exception.AssertError test(6): unittest failure


Btw I think this might be an accident, in mars.c there's this 
code:

     if (global.params.release)
     {   global.params.useInvariants = 0;
         global.params.useIn = 0;
         global.params.useOut = 0;
         global.params.useAssert = 0;  // <-- note
         global.params.useArrayBounds = 1;
         global.params.useSwitchError = 0;
     }

However then it's followed by this code:

     if (global.params.useUnitTests)
         global.params.useAssert = 1;

So it's switched off and then on again. I can't tell if this was 
deliberate or not. If it's on purpose it should be documented.