httpd-test-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Stas Bekman <s...@stason.org>
Subject Re: [patch] hint why the test was skipped
Date Thu, 18 Oct 2001 17:10:15 GMT
William A. Rowe, Jr. wrote:

> From: "Stas Bekman" <stas@stason.org>
> Sent: Wednesday, October 17, 2001 11:14 PM
> 
> 
> 
>>William A. Rowe, Jr. wrote:
>>
>>
>>>From: "Stas Bekman" <stas@stason.org>
>>>Sent: Wednesday, October 17, 2001 10:40 PM
>>>
>>>
>>>> exec ./t/TEST protocol/eliza
>>>>protocol/eliza....skipped: missing Chatbot::Eliza1
>>>>All tests successful, 1 test skipped.
>>>>
>>>>The extension of Test::plan is already somewhat tricky, so in order to 
>>>>preserve the trick we have to make sure that our extension always 
>>>>returns a zero or one value. Since CODE and ARRAY refs were already 
>>>>taken I've used the available HASH ref. Now have module returns a reason 
>>>>for its failure, the CODE and ARRAY refs do or can do the same, here is 
>>>>how it works (this is incomplete, just to see if you like it):
>>>>
>>>>
>>>Without groking the code, definately +1 to allow Foo::try syntax, since we
>>>on non-unix platforms already have problems with Foo/try vs Foo\try ;)
>>>
>>Hmm, I guess you are talking about somethins else. What I try to do now 
>>is to print the reason for the skipped test.
>>
> 
> That is, in and of itself, a great thing!

:)


 
>>Can you please expand on the problem you are talking about? And how do 
>>you want it to be handled?
>>
> 
> Right now, we pass the argument to t/TEST somesuite/item.
> 
> If we adopted the Perl syntax of somesuite::item, the parsers wouldn't have
> to deal with differences between somesuite/item.t and somesuite\item.t 
> which would make documenting the 'invoking' feature much simpler for all
> platforms.
> 
> Right now, we generally worry about the 'unixish' family, and OS2, Win32
> and Netware all accept foo/bar even though their native tounge is foo\bar.
> 
> If we ever support far stranger platforms [and you seem ready with all of
> the file::spec code in the tree ;-] then this will become more difficult.


Apache::Test is built on top of Test::Harness which is used by 'make 
test' of Perl itself on all platforms Perl is running on. And I think 
Perl is running on about the same platforms as Apache, if not more (I'm 
not sure). Therefore I think what's good for Test::Harness is good for 
Apache::Test. And if something is not working the fix should be done in 
Test::Harness.

The only nuance is that Apache::Test tests for the test file existance 
if you explicitly specify these, but since the path is preserved I don't 
think why there should be any problem at all. So if you call t/TEST with 
foo\\bar, Perl will test for -e foo\\bar, we just need to make sure to 
use File::Spec to correctly prepend the t/ prefix. but this is an easy 
fix. Is that the only problem with the test names?

Your suggestion to adopt foo::bar seems to be confusing, since many 
times I'm using file completion and do ./t/TEST t/foo/bar.t using the 
real path, with your proposal you cannot do this anymore, not talking 
about the fact that some tests are just grouped in the same dir, but 
have nothing to do with dir:: prefix. ... I don't know, I suggest to 
solve the problem if we have any and continue using the real paths. What 
do you think?



_____________________________________________________________________
Stas Bekman             JAm_pH      --   Just Another mod_perl Hacker
http://stason.org/      mod_perl Guide   http://perl.apache.org/guide
mailto:stas@stason.org  http://ticketmaster.com http://apacheweek.com
http://singlesheaven.com http://perl.apache.org http://perlmonth.com/


Mime
View raw message