"TAPlug"s for TAP::Harness

Eric Wilhelm scratchcomputing at gmail.com
Wed Jun 10 20:52:39 GMT 2009


# from David Golden
# on Wednesday 10 June 2009 12:57:

>The problem continues to be that "everyone" would need to have
>TAP::Harness and an upgraded toolchain that is TAP::Harness aware.  So
>until configure_requires is widespread, I think you're stuck
>supporting the legacy interface.

No.  Why?  Who is "everyone"?  Anything an author puts in Build.PL or a 
custom build subclass or whatever which demands a TAP::Harness feature 
to simply "make && make test && make install" is either:

    1.  misguided
Or  2.  sufficiently advanced to demand configure_requires &c

It seems that whether we have Test::Harness is then far beyond moot. 
Anything involving Test::Harness in either case will only make matters 
worse.

And if you're talking about a consumer doing something special while you 
run the standard process (e.g. cpantesters), well... you're doing 
something special -- so you install some special stuff and do it.

It might require adding code to M::I, EU::MM, and M::B to inflict 
specialness on unwitting distributions.  (And if anybody replaced 
ACTION_test(), too bad.)

If what you want is to do evil testery things, maybe a 98% solution of 
running `prove` with some options beats a 99% solution of hacking M::I, 
EU::MM, & M::B.

Maybe M::B later defaults to use TAP::Harness - then your Test::Harness 
knobs are broken -- but maybe ACTION_test() is just using the modem to 
order pizza and hookers instead of running tests anyway.  On the "just 
run prove" side, you miss some possible setup and teardown code as well 
as the recursive_test_files bit.

>The related problem is that you have two types of users:
>
>(a) developers, who may or may not do something custom with the
>harness in the build tools of a distribution
>
>(b) end-users, who may or may not want some custom harness behavior
>and likely do not want to mess with the build-tools in an distribution

For (a), see 1 or 2 above.  For (b), I still think you're talking about 
situations where people want features and should therefore pay for them 
(by installing the free upgrade.)

--Eric
-- 
Speak softly and carry a big carrot.
---------------------------------------------------
    http://scratchcomputing.com
---------------------------------------------------


More information about the tapx-dev mailing list