Re: BC, testing, BBC
| From: | bertrand Gugger | Date: | Fri, 19 Nov 2004 19:58:32 +0000 |
| Subject: | Re: BC, testing, BBC | ||
| References: | 1 2 3 4 | Groups: | php.pear.qa |
| Request: | Send a blank email to pear-qa+get-1981@lists.php.net to get a copy of this message | ||
Justin Patrin wrote:
Again, I don't understand your point. Of course tests are meant to test BC (as well as just testing correct functioning). BC is based on previous correct functioning.
BC is never supposed to be broken except in very few cisrumstances (huge security hole, major version # update, etc.). "except" means it can occurs
When the package breaks BC on purpose, yes the tests have to eb changed. So people or machines changing the test should announce it (that's my enquiry).
When new functionality is added, tests should be added. Same as just borned functionality
What are you trying to say? What is this discussion supposed to be about? Try to make the testing procedures more robust.Keep using PEAR, because BC, and when for important reason no more BC, have a way to understand the break in backwards compatibility (BBC), so I can adapt tests from mine and the applications after.
Do you have a problem with some of the current tests? Some packages have no public ones..phpt scripts are quite unknown as you know (refer http://news.php.net/php.pear.dev/34441 )
Are you trying to start a testing RFC? could appreciate, but never: then is it a back list, frozen as here.Actually, Justin, that could lead to more procedural testing and force to explain BBC (what should never give) à+ bertrand Gugger (toggg)