develooper Front page | perl.macperl.porters | Postings from August 2009

P.S.: Could somebody please test this in order for me to fix a bug in Bit::Vector? (update)

Thread Next
From:
O. STeffen BEYer
Date:
August 25, 2009 01:02
Subject:
P.S.: Could somebody please test this in order for me to fix a bug in Bit::Vector? (update)
Message ID:
f4dd31f70908240536o769e2ae1l1d17fbbdc64017fb@mail.gmail.com
P.S.: Anyone with Perl v5.8.6 on a PowerPC G5 Mac Os X v10.4.11?

The original bug report reads:

"I tried to install your module Bit::Vector 6.6 (perl v5.8.6 on a PowerPC
G5 Mac Os X v10.4.11) but it failed the test '28___chunklist.t'.

I ran the test outside of Test::Harness, and found it is failing on
the final value of $bits (=33) at $n=97 and $n=99."

The number of bits he indicated is corroborated by the number of tests,
which is 3 * bits. Normally there would only be 96 tests on a 32 bit
machine.

As far as I can remember, I already heard in the past that the Mac was
simulating on more bit when doing bit shift operations (or always regarded
the carry as a additional bit).
Or am I hallucinating this memory?

Paul Schinder just ran the test scripts I sent and everything went fine.

So the results may depend on the version of MacOS or the compiler used.

Thanks a lot for your support!

Best regards,
Steffen


2009/8/24 O. STeffen BEYer <ostbey@gmail.com>

> Dear MacPerl Porters,
>
> could somebody of you please run the two little scripts "bits.pl" and
> "bits.c" (compile first of course) attached here on a Mac with MacOS X
> (Darwin)?
>
> Somebody reported that two tests in "t/28___chunklist.t" failed on his
> Darwin machine; it turned out that BitVector_Boot() had determined the type
> "unsigned long" to have a size of 33 bits, which is of course odd.
> Apparently 33 bits is not correct and consequently, bits get lost when the
> module expects 33 bits to be passed in each chunk of data.
> I suspect the routine to count the number of bits is somehow fooled under
> Darwin and would like to find out why it was so, and how it should be
> calculated correctly.
>
> Please send back the output (STDOUT) of both bits.pl and bits.c to me -
> thank you very much in advance!
>
> You can also test the attached release candidate Bit::Vector 7.0 to see if
> it passes all tests.
>
> I really would like to release this candidate as soon as possible, because
> I have two other modules depending on it with bugfixes waiting to be
> released as well.
>
> Thank you for any help!
>
> Best regards,
> Steffen Beyer
>

Thread Next


nntp.perl.org: Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at ask@perl.org | Group listing | About