Bug #23972 for Number-WithError: The test suite is failing all over the place

Sun Dec 17 12:10:55 2006 jpierce [...] cpan.org - Ticket created

Subject:

The test suite is failing all over the place

Message body is not shown because it is too large.

Mon Dec 18 05:50:20 2006 smueller [...] cpan.org - Correspondence added

On Sun Dec 17 12:10:55 2006, JPIERCE wrote: Show quoted text

> 2 logs here, one for 5.8.0, one for 5.8.4

[snip long logs] This must be close to the weirdest bug I have yet (not) found. The tests aren't failing any more (see below), but I'm unsure why. Thus, I'll keep the ticket open (or well, rather stalled). Diagnostics as follows: - Depending on whether load Math::Symbolic somewhere, it surfaces or not. (If loaded, it *doesn't* surface.) --> Seems like a hidden dependency, but for the life of me, I can't find it! And that happens even though I wrote all pieces of software involved. - I refactored the constructor tests a little and it stopped breaking. No functional changes, just shuffled the code around a little bit. - The failures were all in the tests which relate to bignum. --> But I don't think it's actually related to bignum support. ... Anyhow. I wasn't expecting the constructor tests to break depending on whether M::S is loaded and thus used the "The test suite is failing all over the place" exclamation refering to the missing my_tan function. That actually just broke one test down the road. Currently, I really can't spend many hours on making the whole thing more elegant (i.e. auto-generate the error propagation formulas with Math::Symbolic), so I opted for reducing the dependencies as you requested. Number::WithError 0.07 will have the dependencies on Math::Symbolic and friends removed. Cheers, Steffen

Mon Dec 18 05:50:21 2006 The RT System itself - Status changed from 'new' to 'open'

Mon Dec 18 05:50:26 2006 smueller [...] cpan.org - Status changed from 'open' to 'stalled'

Mon Dec 18 07:01:36 2006 smueller [...] cpan.org - Correspondence added

On Mon Dec 18 05:50:20 2006, SMUELLER wrote: Show quoted text

> On Sun Dec 17 12:10:55 2006, JPIERCE wrote:

> > 2 logs here, one for 5.8.0, one for 5.8.4

> [snip long logs] > > This must be close to the weirdest bug I have yet (not) found. The tests > aren't failing any more (see below), but I'm unsure why. Thus, I'll keep > the ticket open (or well, rather stalled).

The tests for Number::WithError::LaTeX were failing due to this as well, so I took some more time to track it down. This was an extremely unobvious problem with Math::BigFloat stringification overloading in the rounding subroutine. Adding a couple of "'s fixed the issue. Expect 0.08 soon. Again, thanks for reporting this! Steffen

Mon Dec 18 07:01:38 2006 The RT System itself - Status changed from 'stalled' to 'open'

Mon Dec 18 07:01:41 2006 smueller [...] cpan.org - Status changed from 'open' to 'resolved'

Mon Dec 18 08:15:46 2006 belg4mit [...] MIT.EDU - Correspondence added

Subject:	Re: [rt.cpan.org #23972] The test suite is failing all over the place
Date:	Mon, 18 Dec 2006 08:12:44 -0500
To:	bug-Number-WithError [...] rt.cpan.org
From:	Jerrad Pierce <belg4mit [...] MIT.EDU>

No problem. It seems changes in Big* hs broken several modules in recent times. -- Free map of local environmental resources: http://CambridgeMA.GreenMap.org -- MOTD on Sweetmorn, the 59th of The Aftermath, in the YOLD 3172: 'scuse me while I tend to how I feel...

Mon Dec 18 08:15:52 2006 The RT System itself - Status changed from 'resolved' to 'open'

Mon Dec 18 10:56:26 2006 smueller [...] cpan.org - Status changed from 'open' to 'resolved'

Mon Dec 18 10:56:28 2006 smueller [...] cpan.org - Severity Important added