view release on metacpan or  search on metacpan

lib/Docs/Site_SVD/Data_SecsPack.pm  view on Meta::CPAN

Adding support for packing and unpacking
floats and flushing out to provide full
support for packing and unpacking all
SEMI E5-94 SECSII numeric formats.

=item Data::SecsPack 0.03

There were the same test software failures and Unit Under Test (UUT) failures
for the following platforms:

 Subject: FAIL Data-SecsPack-0.02 ppc-darwin-thread-multi 7.2.0 
 From: nothingmuch@woobling.org 
 perl5: (revision 5.0 version 8 subversion 3) configuration:
 Platform: osname=darwin, osvers=7.2.0, archname=ppc-darwin-thread-multi

 From: "Thurn, Martin" <martin.thurn@ngc.com> 
 Subject: FAIL Data-SecsPack-0.02 sun4-solaris-thread-multi 2.8
 perl5: (revision 5.0 version 9 subversion 0)
 Platform: osname=solaris, osvers=2.8, archname=sun4-solaris-thread-multi

The test software failures were caused by invalid data from the UUT failure.
The UUT failures caused  non-numeric data for C<pass_fail_tolerance> in
C<SecsPackStress.t>. Corrected C<pass_fail_tolerance> by detecting non-numeric
data and returning a failure. 
Failure of UUT C<unpack_float> caused a C<undef> passed to C<binary2hex>.
Corrected C<binary2hex> by detecting and changing C<undef> to to C<''>.

There were two groups of UUT failures. 

In the first group the test results were as the test results was
two member array of C<Math::BigInt> number hash while the test
was expecting just two simple Perl numbers. 
The numbers in the C<Data::BigInt> hash were correct just in a
different format. For example,

 t/Data/SecsPackStress....# Test 2 got: 'U1[1] 80
 L[4]
   A[0]
   A[5] ARRAY
   L[6]
     A[12] Math::BigInt
     A[4] HASH
     A[4] sign
     A[1] +
     A[5] value
     U1[1] 128
   L[6]
     A[12] Math::BigInt
     A[4] HASH
     A[4] sign
     A[1] +
     A[5] value
     U1[1] 0
 ' (t/Data/SecsPackStress.t at line 184)
 #   Expected: 'U1[1] 80
 U1[2] 128 0
 '

The second group of failures the UUT C<unpack_float> subroutine
is always returning a NaN. This subroutine heavily uses the
native floating point for calcuations and it appears to
be limiting out with C<Not a Number> type error.

 # Test 27 got: '0' (t/Data/SecsPackStress.t at line 396 fail #2)
 #    Expected: '1' (got: NaN, expected: -10.5
 # actual tolerance: NaN, expected tolerance: 0.0001)

Check version of C<Math::BigInt> in the ActiveState Perl 5.06
distribution to those on CPAN. ActiveState C<Math::BigInt>
was 0.01 while CPAN was 1.70. The difference between the two
is night and day. Install C<Math::BigInt> 1.70.
Tests fails the same as above. 

Corrected all C<Data::SecsPack> subroutines to use new
features and intefaces of C<Math::BigInt> 1.70 including
a load C<die> if the C<Math::BitInt> version is not 1.70
or greater

Replaced as many native float operations as possible with
many new C<Math::BigInt> subroutines in Version 1.70 as possible.
Removed work arounds for subroutines such as C<rsft> do not
not work completely in 1.0 but do in 1.70. 
The C<unpack_float> subroutine
does not depend heavily on native float which comes in a different
flavor for each and every site. There may be exceptions to
this statement but if they are, they a few and far between.

=item Data::SecsPack 0.04

The anticipation was that there would be minor differences in 
floating point numbers due to different native floating point
implementations. This concern did not materialize. In addition, 
C<Data::BigFloat> replaced all native floating points.
Since floating point differences are no longer anticipation,
removed all tolerance criteria from the test scripts.
To pass the tests, the results will now have to be exactly
as expected.

Added the formats U8 and S8 to allowable formats for
C<unpack_int>. This was an oversight, a mistake.

Added the C<ascii_float> option to the C<str2float>
subroutine.  This is a new feature.

Switch to C<Data::Startup> for handling options.

=item Data::SecsPack 0.05

The C<str2float> subroutine did not recognize 0 as a number.
Added 0 to tests for C<str2float>.

=item Data::SecsPack 0.06

Changed the required version for C<Math::BigFloat> from 1.40 to 1.39.
Somehow the perquisites have some mysterious behaviors.

Failure:

 From: mhoyt@houston.rr.com 
 Subject: FAIL Tie-Layers-0.04 darwin-thread-multi-2level 7.0 

PERL_DL_NONLAZY=1 /usr/bin/perl "-MExtUtils::Command::MM" "-e" 
"test_harness(0, 'blib/lib', 'blib/arch')" t/Tie/Layers.t
t/Tie/Layers....Math::BigFloat version 1.4 required--this is only version 1.39 

Documentation woes.

lib/Docs/Site_SVD/Data_SecsPack.pm  view on Meta::CPAN

Adding support for packing and unpacking
floats and flushing out to provide full
support for packing and unpacking all
SEMI E5-94 SECSII numeric formats.

\=item Data::SecsPack 0.03

There were the same test software failures and Unit Under Test (UUT) failures
for the following platforms:

 Subject: FAIL Data-SecsPack-0.02 ppc-darwin-thread-multi 7.2.0 
 From: nothingmuch@woobling.org 
 perl5: (revision 5.0 version 8 subversion 3) configuration:
 Platform: osname=darwin, osvers=7.2.0, archname=ppc-darwin-thread-multi

 From: "Thurn, Martin" <martin.thurn@ngc.com> 
 Subject: FAIL Data-SecsPack-0.02 sun4-solaris-thread-multi 2.8
 perl5: (revision 5.0 version 9 subversion 0)
 Platform: osname=solaris, osvers=2.8, archname=sun4-solaris-thread-multi

The test software failures were caused by invalid data from the UUT failure.
The UUT failures caused  non-numeric data for C<pass_fail_tolerance> in
C<SecsPackStress.t>. Corrected C<pass_fail_tolerance> by detecting non-numeric
data and returning a failure. 
Failure of UUT C<unpack_float> caused a C<undef> passed to C<binary2hex>.
Corrected C<binary2hex> by detecting and changing C<undef> to to C<''>.

There were two groups of UUT failures. 

In the first group the test results were as the test results was
two member array of C<Math::BigInt> number hash while the test
was expecting just two simple Perl numbers. 
The numbers in the C<Data::BigInt> hash were correct just in a
different format. For example,

 t/Data/SecsPackStress....# Test 2 got: 'U1[1] 80
 L[4]
   A[0]
   A[5] ARRAY
   L[6]
     A[12] Math::BigInt
     A[4] HASH
     A[4] sign
     A[1] +
     A[5] value
     U1[1] 128
   L[6]
     A[12] Math::BigInt
     A[4] HASH
     A[4] sign
     A[1] +
     A[5] value
     U1[1] 0
 ' (t/Data/SecsPackStress.t at line 184)
 #   Expected: 'U1[1] 80
 U1[2] 128 0
 '

The second group of failures the UUT C<unpack_float> subroutine
is always returning a NaN. This subroutine heavily uses the
native floating point for calcuations and it appears to
be limiting out with C<Not a Number> type error.

 # Test 27 got: '0' (t/Data/SecsPackStress.t at line 396 fail #2)
 #    Expected: '1' (got: NaN, expected: -10.5
 # actual tolerance: NaN, expected tolerance: 0.0001)

Check version of C<Math::BigInt> in the ActiveState Perl 5.06
distribution to those on CPAN. ActiveState C<Math::BigInt>
was 0.01 while CPAN was 1.70. The difference between the two
is night and day. Install C<Math::BigInt> 1.70.
Tests fails the same as above. 

Corrected all C<Data::SecsPack> subroutines to use new
features and intefaces of C<Math::BigInt> 1.70 including
a load C<die> if the C<Math::BitInt> version is not 1.70
or greater

Replaced as many native float operations as possible with
many new C<Math::BigInt> subroutines in Version 1.70 as possible.
Removed work arounds for subroutines such as C<rsft> do not
not work completely in 1.0 but do in 1.70. 
The C<unpack_float> subroutine
does not depend heavily on native float which comes in a different
flavor for each and every site. There may be exceptions to
this statement but if they are, they a few and far between.

\=item Data::SecsPack 0.04

The anticipation was that there would be minor differences in 
floating point numbers due to different native floating point
implementations. This concern did not materialize. In addition, 
C<Data::BigFloat> replaced all native floating points.
Since floating point differences are no longer anticipation,
removed all tolerance criteria from the test scripts.
To pass the tests, the results will now have to be exactly
as expected.

Added the formats U8 and S8 to allowable formats for
C<unpack_int>. This was an oversight, a mistake.

Added the C<ascii_float> option to the C<str2float>
subroutine.  This is a new feature.

Switch to C<Data::Startup> for handling options.

\=item Data::SecsPack 0.05

The C<str2float> subroutine did not recognize 0 as a number.
Added 0 to tests for C<str2float>.

\=item Data::SecsPack 0.06

Changed the required version for C<Math::BigFloat> from 1.40 to 1.39.
Somehow the perquisites have some mysterious behaviors.

Failure:

 From: mhoyt@houston.rr.com 
 Subject: FAIL Tie-Layers-0.04 darwin-thread-multi-2level 7.0 

PERL_DL_NONLAZY=1 /usr/bin/perl "-MExtUtils::Command::MM" "-e" 
"test_harness(0, 'blib/lib', 'blib/arch')" t/Tie/Layers.t
t/Tie/Layers....Math::BigFloat version 1.4 required--this is only version 1.39 

Documentation woes.