From: Brian Ravnsgaard Riis (brian_at_[hidden])
Date: 2005-04-18 04:50:30
Vladimir Prus wrote:
> On Sunday 17 April 2005 22:32, Brian Riis wrote:
>>>Would be nice if some one else could explain why you would ever use run
>>>as opposed to unit-test.
>>Well, now you pointed me in the right direction, I took a look at those
>>rules, and run is quite different from unit-test in invocation. This
>>actually means that run is more powerful, since you can pass arguments
>>to the command to run. For simple unit tests that require no arguments
>>and simply indicate success or failure by return code unit-test is
>>simpler to use.
> There's another important difference between 'unit-test' and 'run'. The 'run'
> rule is meant to work with Boost's regression testing system. It captures the
> program output, and creates some additional files. So, if you'll use 'run'
> outside of Boost regression system, any test failure will produce a lot of
> strange output, and store the actual output from the test in a file. OTOH,
> with unit-test, you'll see program output, followed by the "failed" message.
> So, for humans, 'unit-test' might be more convenient.
That's sorta what I expected from reading through testing.jam as well.
It might be a good idea to state the need to "import testing ;" in the
manual around the place where the unit-test target is described.
> - Volodya
-- /Brian Riis
Boost-Build list run by bdawes at acm.org, david.abrahams at rcn.com, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk