Changes between Initial Version and Version 1 of PerlSixTesting

Show
Ignore:
Timestamp:
12/09/08 21:44:06 (13 years ago)
Author:
moritz
Comment:

First mind map of Perl 6 S24, Testing

Legend:

Unmodified
Added
Removed
Modified
  • PerlSixTesting

    v1 v1  
     1= Purpose, goals = 
     2    * built-in, very easy testing 
     3    * gain exactness 
     4    * good diagnostics 
     5 
     6= Definition = 
     7    * series of discrete operations with boolean result, marked as tests 
     8 
     9= Syntax = 
     10== overall structure == 
     11        - plan(2); tests; # done 
     12        - plan(*); tests; tests-done(); # names subject to discussion ;-) 
     13 
     14== single tests == 
     15      we have two proposal: 
     16 
     17=== adverb on operators === 
     18                $x eq 'foo'     :ok('Test descriptions'); 
     19                $x == 23        :ok('explicit numeric test'); 
     20                ?$x             :ok<stuff>      # used to be ok($x, 'stuff'); 
     21                try { dispatch() } :ok<...>     # used to be lives_Ok { ... }, '...'; 
     22            (it was also proposed to name it :test instead of :ok) 
     23            pros: 
     24                * good diagnose of output 
     25                * force exact comparison semantics 
     26                * no pollution of namespace 
     27            cons: 
     28                * have to auto-generate ops with adverbs 
     29                * more complicated grammar needed 
     30                * the fact that a line contains a test is somewhere in the 
     31                  middle, so not too obvious 
     32 
     33=== normal sub calls, like now === 
     34                ok $x eq 'foo', 'description'; 
     35                dies_ok { ... }, 'description'; 
     36            pros: 
     37                * easier to implement 
     38                * it's been proven to work out well 
     39                * visually tests are very easy to spot 
     40            cons: 
     41                * it's *very* hard to get good diagnostics out of failed tests 
     42                  (advanced macro fiddling needed) 
     43                * clutters up namespace 
     44                * if we don't deprecate is(), we have to find sane comparison 
     45                  semantics, fix up the tests (we have to do that anyway...) 
     46                  and educate our test writes. Currently infix:<eqv> is a 
     47                  candidate (advocated by pmichaud, because it's very dwimmy), 
     48                  infix:<eq> being another (advocated by mority, because it's 
     49                  very simple and easy to explain). 
     50 
     51= Backends and standard backend = 
     52  
     53    * all the test logic should be easily overridable (for example for storing 
     54        test results in a db, or different test output format) 
     55    * simplest idea:  
     56        there's a Test::Backend object in $*TEST_BACKEND, and plan() class 
     57        the .plan method, :ok('...') adverbs on ops call. 
     58    * the default backend will emit TAP output 
     59    * the default backend will consider a test run failed if any single test 
     60      failed, or in case of plan mismatch