New release: Math::GSL 0.13_03

[Jonathan Leto]

Howdy,

New dev release just got sent to CPAN, you can download *right now*
temporarily here:

http://leto.net/tmp/Math-GSL-0.13_03.tar.gz

Check out http://leto.net/code/Math-GSL/ for more details.

Cheers,

--
[---------------------]
Jonathan Leto
jal...@gmail.com

[Sisyphus]

----- Original Message -----
From: "Jonathan Leto" <jal...@gmail.com>

> New dev release just got sent to CPAN, you can download *right now*
> temporarily here:
>
> http://leto.net/tmp/Math-GSL-0.13_03.tar.gz
>

Hi Jonathan,
As regards my 64-bit linux build the only failing tests are now in SF.t.
Here's how the five failures show up in the'./Build test' output:

---------------------------------------------------
# Failed test 'gsl_sf_atanint_e(1.0e-10,$r) ?= 1.00000000000000004e-10,
# res= +-3.64138193850019049e-27, eps=0'
# at t/SF.t line 1148.
# (in Math::GSL::SF::Test->TEST_THE_KITCHEN_SINK)

# Failed test 'gsl_sf_expint_3_e(1.0e-10,$r) ?= 1.00000000000000004e-10,
# res= +-3.64138193850019049e-27, eps=0'
# at t/SF.t line 1148.
# (in Math::GSL::SF::Test->TEST_THE_KITCHEN_SINK)

# Failed test 'gsl_sf_dawson_e(1.0e-15,$r) ?= 1.00000000000000008e-15,
# res= +-7.77112732560249667e-32, eps=0'
# at t/SF.t line 1148.
# (in Math::GSL::SF::Test->TEST_THE_KITCHEN_SINK)

# Failed test 'gsl_sf_fermi_dirac_m1_e( 10.0,$r) ?= 0.99995460213129761,
# res= +-4.401860820291148e-17, eps=4.43555616580345222e-18'
# at t/SF.t line 1148.
# (in Math::GSL::SF::Test->TEST_THE_KITCHEN_SINK)

# Failed test 'gsl_sf_Si_e(1.0e-10,$r) ?= 1.00000000000000004e-10,
# res= +-3.64138193850019049e-27, eps=0'
# at t/SF.t line 1148.
# (in Math::GSL::SF::Test->TEST_THE_KITCHEN_SINK)
# Looks like you failed 5 tests of 1108.
---------------------------------------------------

I'm assuming that's all you'll need to see in order to easily fix the
failures - but please let me know if that assumption is incorrect :-)

Cheers,
Rob

[Jonathan Leto]

Howdy,

Could you send the output of those failing test when you have the
DEBUG environment
variable set? It will give me a little more info to debug with.

Something like this:

DEBUG=1 ./Build test --verbose 1 --test_files t/SF.t

I think I can easily fix the failing tests which have eps=0, the other
failing test looks like possibly
a loss of precision by GSL.

Cheers,

--
[---------------------]
Jonathan Leto
jal...@gmail.com

[Sisyphus]

----- Original Message -----
From: "Jonathan Leto" <jal...@gmail.com>
To: <math-g...@googlegroups.com>
Sent: Monday, October 20, 2008 1:27 PM
Subject: Re: New release: Math::GSL 0.13_03

>
> Howdy,
>
> Could you send the output of those failing test when you have the
> DEBUG environment
> variable set? It will give me a little more info to debug with.
>
> Something like this:
>
> DEBUG=1 ./Build test --verbose 1 --test_files t/SF.t
>

No problem:

not ok 70 - gsl_sf_atanint_e(1.0e-10,$r) ?= 1.00000000000000004e-10,
# res= +-3.64138193850019049e-27, eps=0
result->err: 7.08278239572255227e-15
result->val: -1.89511781635593746
got gsl_sf_expint_E1_e(-1.0,$r) = -1.89511781635593746
expected : -1.89511781635593676
difference : 7.01261965163624268e-16

not ok 272 - gsl_sf_expint_3_e(1.0e-10,$r) ?= 1.00000000000000004e-10,
# res= +-3.64138193850019049e-27, eps=0
result->err: 4.79348010992590794e-16
result->val: 0.0915633339397880913
got gsl_sf_expint_E1_scaled_e(10.0,$r) = 0.0915633339397880913
expected : 0.0915633339397880819
difference : 9.43933516420192298e-18

not ok 315 - gsl_sf_dawson_e(1.0e-15,$r) ?= 1.00000000000000008e-15,
# res= +-7.77112732560249667e-32, eps=0
result->err: 4.71809514011506086e-15
result->val: 1.06242057574603677
got gsl_sf_exprel_n_e(50, 3.0,$r) = 1.06242057574603677
expected : 1.06242057574603683
difference : 6.46184494801360643e-17

not ok 437 - gsl_sf_fermi_dirac_m1_e( 10.0,$r) ?= 0.99995460213129761,
# res= +-4.401860820291148e-17, eps=4.43555616580345222e-18
result->err: 8.85829810396271826e-16
result->val: 0.797884560802865406
got gsl_sf_hazard_e( 0.0,$r) = 0.797884560802865406
expected : 0.797884560802865356
difference : 4.98190898257089287e-17

not ok 996 - gsl_sf_Si_e(1.0e-10,$r) ?= 1.00000000000000004e-10,
# res= +-3.64138193850019049e-27, eps=0

# Failed test 'gsl_sf_Si_e(1.0e-10,$r) ?= 1.00000000000000004e-10,
# res= +-3.64138193850019049e-27, eps=0'
# at t/SF.t line 1148.
# (in Math::GSL::SF::Test->TEST_THE_KITCHEN_SINK)

result->err: 4.83605833232409254e-15
result->val: 0.544491762584919159
got gsl_sf_exprel_n_e(3, -3.0,$r) = 0.544491762584919159
expected : 0.544491762584919124
difference : 3.51281503885303437e-17

Cheers,
Rob