|
Boost : |
From: Beman Dawes (beman_at_[hidden])
Date: 2000-06-20 09:27:22
At 08:31 AM 6/19/2000 -0500, Ed Brey wrote:
>From: "Beman Dawes" <beman_at_[hidden]>
>>
>> Third question for each compiler is how do you tell if a compilation
>has
>> errors? For VC++ I am grepping the output for "error" for lack of a
>better
>> way.
>
>A solution to this problem is a more complete test. Rather than
>looking for compiler errors, the script can try to run the compiled
>test harness and then compare the output against a known good output.
>If the compiler thinks there are errors in the library, the test will
>fail since there won't be an executable of the test harness.
>
>Testing run-time output is something we should do anyway, given that
>we have experience with compilers occasionally accepting code and then
>generating incorrect binaries. The amount of work required should be
>minimal, since the thoroughness of the test harness required to test
>the library should also adequately test the compiler.
Yes, agreed. But one step at a time. Enough trouble just getting compiles
to work. Also see my reply to Lois's message.
--Beman
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk