FAQ
Hello

It looks like the original developer of Compress::Bzip2 has disappeared:

The original message was received at Sun, 20 Jul 2003 14:51:31 -0400 (EDT)
from ifi.informatik.uni-stuttgart.de [129.69.211.1]

----- The following addresses had permanent fatal errors -----
(reason: 550 5.1.1 User unknown)
azemgi@rupert
(reason: 550 5.1.1 User unknown)
(expanded from: <azemgi@rupert.informatik.uni-stuttgart.de>)

But then, I've noticed that on www.cpan.org, a Version 1.02 of
Compress::Bzip2 can be found. But why does the old CPAN shell
download version 1.00 ? I'm using the mirror on:
ftp://sunsite.cnlab-switch.ch/mirror/CPAN - the version 1.02 is even
there, in KCARNUT's directory. It is merely not found by the CPAN
shell.

Marco Carnut's new version already solves the BZ2_ prefix problem.
But it does still not solve the failing test case!

My questions:
- How comes the wrong download? CPAN shell bug or did Marco Carnut
something wrong?
- Shouldn't it be made a requirement for CPAN uploads to pass 'make test'?

Cheers
Christian.
Date: Sun, 20 Jul 2003 20:51:20 +0200
To: azemgi@rupert.informatik.uni-stuttgart.de
From: Christian Jaeger <christian.jaeger@ethlife.ethz.ch>
Subject: [PATCH] making Compress::Bzip2 work

Hello

I had to patch both the .xs and the t/test.t files to make it work
and install ok on a Debian Woody box.

Please review the attached diff and consider uploading a new version to CPAN.

Thanks,
Christian Jaeger

Search Discussions

  • Vadim Konovalov at Jul 20, 2003 at 7:49 pm
    Why you use such a strange version in your patch:


    -$VERSION = "1.00";
    +$VERSION = "1.01cj";

    Please don't do that, you'll mess things a lot.

    Besides, you're trying to patch against version 1.00 and you're claiming
    that there exist version 1.02. Isn't this strange?

    Yet another besides, best solution will be to succort both versions of
    BZ-library using set of #define's and not unconditionally replace elder
    functiopn calls with newer ones.

    Best regards,
    Vadim.


    ----- Original Message -----
    From: "Christian Jaeger" <christian.jaeger@ethlife.ethz.ch>
    To: <perl5-porters@perl.org>
    Cc: "Marco Carnut" <kiko@tempest.com.br>
    Sent: Sunday, July 20, 2003 11:15 PM
    Subject: Fwd: [PATCH] making Compress::Bzip2 work

    Hello

    It looks like the original developer of Compress::Bzip2 has disappeared:

    The original message was received at Sun, 20 Jul 2003 14:51:31 -0400 (EDT)
    from ifi.informatik.uni-stuttgart.de [129.69.211.1]

    ----- The following addresses had permanent fatal errors -----
    (reason: 550 5.1.1 User unknown)
    azemgi@rupert
    (reason: 550 5.1.1 User unknown)
    (expanded from: <azemgi@rupert.informatik.uni-stuttgart.de>)

    But then, I've noticed that on www.cpan.org, a Version 1.02 of
    Compress::Bzip2 can be found. But why does the old CPAN shell
    download version 1.00 ? I'm using the mirror on:
    ftp://sunsite.cnlab-switch.ch/mirror/CPAN - the version 1.02 is even
    there, in KCARNUT's directory. It is merely not found by the CPAN
    shell.

    Marco Carnut's new version already solves the BZ2_ prefix problem.
    But it does still not solve the failing test case!

    My questions:
    - How comes the wrong download? CPAN shell bug or did Marco Carnut
    something wrong?
    - Shouldn't it be made a requirement for CPAN uploads to pass 'make test'?

    Cheers
    Christian.
    Date: Sun, 20 Jul 2003 20:51:20 +0200
    To: azemgi@rupert.informatik.uni-stuttgart.de
    From: Christian Jaeger <christian.jaeger@ethlife.ethz.ch>
    Subject: [PATCH] making Compress::Bzip2 work

    Hello

    I had to patch both the .xs and the t/test.t files to make it work
    and install ok on a Debian Woody box.

    Please review the attached diff and consider uploading a new version to
    CPAN.
    Thanks,
    Christian Jaeger
  • Christian Jaeger at Jul 20, 2003 at 8:25 pm

    At 23:50 Uhr +0400 20.07.2003, Vadim Konovalov wrote:
    Why you use such a strange version in your patch:


    -$VERSION = "1.00";
    +$VERSION = "1.01cj";

    Please don't do that, you'll mess things a lot.
    Ok sorry. Well my patch was not thought to be applied unchanged.
    Besides, you're trying to patch against version 1.00 and you're claiming
    that there exist version 1.02. Isn't this strange?
    I didn't know there was 1.02 at the time I wrote the patch. I was too
    lazy to write everything again - if I wanted, I'd write a more
    complete interface anyway.
    Yet another besides, best solution will be to succort both versions of
    BZ-library using set of #define's and not unconditionally replace elder
    functiopn calls with newer ones.
    I'm not the maintainer, I just wanted to tell what worked for me.

    If anyone is serious about the Compress::Bzip2 module, he should also
    interface the other methods of libbz2, not only the two buffer
    oriented that are supported now.

    Christian.
  • Christian Jaeger at Jul 20, 2003 at 11:37 pm

    At 22:25 Uhr +0200 20.07.2003, I wrote:
    At 23:50 Uhr +0400 20.07.2003, Vadim Konovalov wrote:
    Yet another besides, best solution will be to succort both versions of
    BZ-library using set of #define's and not unconditionally replace elder
    functiopn calls with newer ones.
    I'm not the maintainer, I just wanted to tell what worked for me.
    Ok I've now written a new version 1.03 which checks if the prefix is needed.
    See: http://pflanze.mine.nu/~chris/scratch/Bzip2/
    If anyone is serious about the Compress::Bzip2 module, he should
    also interface the other methods of libbz2, not only the two buffer
    oriented that are supported now.
    Not yet done.

    Christian.
  • Vadim Konovalov at Jul 21, 2003 at 4:40 am

    I'm not the maintainer, I just wanted to tell what worked for me.
    Ok I've now written a new version 1.03 which checks if the prefix is needed.
    See: http://pflanze.mine.nu/~chris/scratch/Bzip2/
    I beleive it's much better now :)
    Thanks for efforts, any improvement to CPAN module is good.

    I am not 100% sure but it seems to me that p5p list is not the right place
    to discuss such things, but something like modules@cpan.org, but I am not
    sure.

    Could you please anyone advice which is the right place to discuss CPAN
    modules and how should I subscribe there?

    Best regards,
    Vadim.
  • Andreas J Koenig at Jul 21, 2003 at 7:17 am

    "Vadim Konovalov" <vkonovalov@peterstar.ru> writes:

    to discuss such things, but something like modules@cpan.org, but I am not
    sure.
    modules@perl.org is more about administrative issues on CPAN.
    Everything around user registration, namespace registration, change of
    ownership, etc. that cannot be handled via PAUSE web interface
    directly. It's not really a mailing list as it cannot be subscribed
    to.
    Could you please anyone advice which is the right place to discuss CPAN
    modules and how should I subscribe there?
    module-authors@perl.org. Subscribe by writing to
    module-authors-subscribe@perl.org

    --
    andreas
  • Michael G Schwern at Jul 21, 2003 at 7:44 pm

    On Mon, Jul 21, 2003 at 08:39:26AM +0400, Vadim Konovalov wrote:
    I am not 100% sure but it seems to me that p5p list is not the right place
    to discuss such things, but something like modules@cpan.org, but I am not
    sure.

    Could you please anyone advice which is the right place to discuss CPAN
    modules and how should I subscribe there?
    modules@perl.org for CPAN administration issues like module ownership.
    module-authors@perl.org is a handy list for module authors talking to other
    module authors.


    --
    They had applied the blinders of steam-grilled hamburgers to my eyes.
  • Vadim Konovalov at Jul 21, 2003 at 4:40 am

    -$VERSION = "1.00";
    +$VERSION = "1.01cj";

    Please don't do that, you'll mess things a lot.
    Ok sorry. Well my patch was not thought to be applied unchanged.
    In this case you should clearly state so, with few exclamation marks
    included!

    Vadim.
  • Michael G Schwern at Jul 20, 2003 at 9:45 pm

    On Sun, Jul 20, 2003 at 09:15:56PM +0200, Christian Jaeger wrote:
    But then, I've noticed that on www.cpan.org, a Version 1.02 of
    Compress::Bzip2 can be found. But why does the old CPAN shell
    download version 1.00 ?
    Because that's what the CPAN/modules/02packages.details.txt.gz module
    index says is the latest version of Compress::Bzip2 is. Its a CPAN indexing
    issue. The CPAN shell is working correctly. But its not a bug...

    I'm using the mirror on:
    ftp://sunsite.cnlab-switch.ch/mirror/CPAN - the version 1.02 is even
    there, in KCARNUT's directory. It is merely not found by the CPAN
    shell.
    This is because KCARNUT is not the owner of the Compress::Bzip2 namespace.
    Neither is ZOOVY who uploaded a 1.01 version. Since CPAN uses a
    first-come first-serve heuristic for namespace ownership by default, AZEMGI
    is the owner of the namespace and his is the one that gets indexed.
    search.cpan.org says otherwise but it doesn't hold itself to only the
    index.

    Now, since the last upload from AZEMGI was in 1999 and the last upload from
    ZOOVY was 2001, KCARNUT has a pretty strong case for taking over the
    namespace. He should contact modules@perl.org, azemgi@cpan.org and
    zoovy@cpan.org, show that the two previous authors are AWOL, that the
    current versions are broken and ask that ownership of the module be turned
    over to him. I've CC'd all the authors.

    - Shouldn't it be made a requirement for CPAN uploads to pass 'make test'?
    No. There are logistical problems which make this impossible. Its also
    against the CPAN philosophy of being an open archive. If you want to know
    what works and what doesn't, testers.cpan.org will tell you. You can even
    build your own alternative CPAN index based on that information if you like.

    The logistical problems are quite fundamental. How does CPAN know to get
    the libbz2 shared library to make Compress::Bzip2 work? Who's going to
    get it an install it even if it could figure that out? What about all the
    non-Unix modules, how do we test them? OS specific modules? Database
    specific modules? Modules which require a certain version of Perl?
    Certain special patches to Perl?

    And then there are some slippery slope issues. What if the module has no
    tests? What if it has one test? What if it has a 1000 tests but its just
    print "ok $_" for 1..1000?

    Something best left to a side project.


    --
    Kindly do not attempt to cloud the issue with facts.
  • Brian Horakh at Jul 21, 2003 at 3:07 am
    To all:

    I gave up in frustration, I've had this patched for over two years and i've simply been giving the patched version to anybody who's asked.
    There's actually two bugs in the version on CPAN right now:
    The first is that it doesn't work with the current libbzip2
    The second is that it doesn't support a variable size for a decompression buffer, so if you have a compressed file which has a more than a 20:1 ratio it simply returns undef and your stuck. It's worth mentioning the bzip2 algorithim regularly achieves better than 100:1 when dealing with highly repetitive XML data.

    Despite multiple attempts, and lots of unanswered questions I couldn't get the CPAN testers to pass a Compress::Bzip2 because nobody seemed to have libbzip installed.
    If somebody wants to give me a hand here i'd sure appreciate it. (Also if KCARNUT wants to take it over, then thats fine by me too)

    I have the most current copy with my patches on my machine at work and i'd be happy to send it to you.



    Brian A. Horakh
    Chief Technical Guy
    Zoovy, Inc.
    T: 1-877-966-8948 ext 111
    F: 425-699-7849

    Michael G Schwern <schwern@pobox.com> 07/20/03 02:45PM >>>
    On Sun, Jul 20, 2003 at 09:15:56PM +0200, Christian Jaeger wrote:
    But then, I've noticed that on www.cpan.org, a Version 1.02 of
    Compress::Bzip2 can be found. But why does the old CPAN shell
    download version 1.00 ?
    Because that's what the CPAN/modules/02packages.details.txt.gz module
    index says is the latest version of Compress::Bzip2 is. Its a CPAN indexing
    issue. The CPAN shell is working correctly. But its not a bug...

    I'm using the mirror on:
    ftp://sunsite.cnlab-switch.ch/mirror/CPAN - the version 1.02 is even
    there, in KCARNUT's directory. It is merely not found by the CPAN
    shell.
    This is because KCARNUT is not the owner of the Compress::Bzip2 namespace.
    Neither is ZOOVY who uploaded a 1.01 version. Since CPAN uses a
    first-come first-serve heuristic for namespace ownership by default, AZEMGI
    is the owner of the namespace and his is the one that gets indexed.
    search.cpan.org says otherwise but it doesn't hold itself to only the
    index.

    Now, since the last upload from AZEMGI was in 1999 and the last upload from
    ZOOVY was 2001, KCARNUT has a pretty strong case for taking over the
    namespace. He should contact modules@perl.org, azemgi@cpan.org and
    zoovy@cpan.org, show that the two previous authors are AWOL, that the
    current versions are broken and ask that ownership of the module be turned
    over to him. I've CC'd all the authors.

    - Shouldn't it be made a requirement for CPAN uploads to pass 'make test'?
    No. There are logistical problems which make this impossible. Its also
    against the CPAN philosophy of being an open archive. If you want to know
    what works and what doesn't, testers.cpan.org will tell you. You can even
    build your own alternative CPAN index based on that information if you like.

    The logistical problems are quite fundamental. How does CPAN know to get
    the libbz2 shared library to make Compress::Bzip2 work? Who's going to
    get it an install it even if it could figure that out? What about all the
    non-Unix modules, how do we test them? OS specific modules? Database
    specific modules? Modules which require a certain version of Perl?
    Certain special patches to Perl?

    And then there are some slippery slope issues. What if the module has no
    tests? What if it has one test? What if it has a 1000 tests but its just
    print "ok $_" for 1..1000?

    Something best left to a side project.


    --
    Kindly do not attempt to cloud the issue with facts.
  • Marco Carnut at Jul 21, 2003 at 4:43 pm
    Folks,

    Wow... a single weekend without checking my email and so much
    has happened... :]

    Just to clarify a few things:
    This is because KCARNUT is not the owner of the Compress::Bzip2
    namespace. Neither is ZOOVY who uploaded a 1.01 version. Since
    CPAN uses a first-come first-serve heuristic for namespace
    ownership by default, AZEMGI is the owner of the namespace and
    his is the one that gets indexed. search.cpan.org says otherwise
    but it doesn't hold itself to only the index.

    Now, since the last upload from AZEMGI was in 1999 and the
    last upload from ZOOVY was 2001, KCARNUT has a pretty strong case
    for taking over the namespace. He should contact modules@perl.org,
    azemgi@cpan.org and zoovy@cpan.org, show that the two previous authors
    are AWOL, that the current versions are broken and ask that ownership
    of the module be turned over to him. I've CC'd all the authors.
    I tried contacting the original author but to send my patch and, like
    most of you, received no answer. After a few days with some friends
    asking me for the BZ2_-prefix corrected version, I enrolled myself on
    PAUSE and shared it anyway to see if it worked. Sorry if I did something
    wrong in my PAUSE-newbie condition. I even got a message telling me I
    wasn't the owner of the namespace, and another telling that it failed
    testing on some systems, so I was surprised when it showed up in the
    Recent page.

    I'm not really sure if I'm the best one to take over the namespace,
    although I surely can do that if no one else more knowledgeable on
    libbzip2 wants it.
    No. There are logistical problems which make this impossible.
    (...)
    Learned that in practice. :)

    Now that Christian Jaeger has done it the right way in his 1.03 version,
    what shall we do?

    BTW, I did something similar with Net::Pcap -- I needed the
    "findalldevs()"
    function, which wasn't present in 0.4; added it myself; contacted the
    author
    but received no reply; then I posted it through PAUSE. Again, if no one
    else wants the namespace, I can maintain it.

    --Marco "Kiko" Carnut, CISSP
    --Tempest Security Technologies -- www.tempest.com.br

Related Discussions

Discussion Navigation
viewthread | post
Discussion Overview
groupperl5-porters @
categoriesperl
postedJul 20, '03 at 7:16p
activeJul 21, '03 at 7:44p
posts11
users6
websiteperl.org

People

Translate

site design / logo © 2021 Grokbase