OpenPKG CVS Repository
  http://cvs.openpkg.org/
  ____________________________________________________________________________

  Server: cvs.openpkg.org                  Name:   Ralf S. Engelschall
  Root:   /e/openpkg/cvs                   Email:  [EMAIL PROTECTED]
  Module: openpkg-web                      Date:   29-Jul-2003 20:52:12
  Branch: HEAD                             Handle: 2003072919521200

  Modified files:
    openpkg-web             faq.wml

  Log:
    one entry obsoleted by OpenPKG 1.3, two new added for answering new
    things

  Summary:
    Revision    Changes     Path
    1.39        +63 -32     openpkg-web/faq.wml
  ____________________________________________________________________________

  patch -p0 <<'@@ .'
  Index: openpkg-web/faq.wml
  ============================================================================
  $ cvs diff -u -r1.38 -r1.39 faq.wml
  --- openpkg-web/faq.wml       12 Jul 2003 19:38:24 -0000      1.38
  +++ openpkg-web/faq.wml       29 Jul 2003 18:52:12 -0000      1.39
  @@ -430,38 +430,6 @@
       </ol>
   </faq>
   
  -<faq id="why-compress"
  -     title="Why does/did the OpenPKG bootstrap use \"compress\" data format?">
  -    The "<tt>openpkg*.src.sh</tt>" source bootstrap should run on as
  -    many platforms as possible. All platforms within the scope of
  -    OpenPKG support unpacking of data that comes in "compress" format.
  -    This is true for very old but still in production UNIX systems. It
  -    is also true for very new LINUX systems which do not have an
  -    uncompress(1) tool installed by default but come with a gunzip(1)
  -    which can unpack "compress" format and is detected and used by the
  -    bootstrap. There is no other packed data format with equal
  -    availability.
  -    <p>
  -    When a user creates a
  -    "<tt>openpkg-*.<i>arch</i>-<i>os</i>-<i>hierarchy</i>.sh</tt>"
  -    binary bootstrap the resulting script was in "compress" format in
  -    OpenPKG v1.0, v1.1 and CURRENT until 20030110.  This required the
  -    availability of compress(1) to the user.  As described above, the
  -    latest incarnations of LINUX omit that crucial tool and gzip(1) is
  -    not a full featured replacement as it cannot create "compress"
  -    format.  For this reason we dropped compression for binary bootstrap
  -    entirely beginning with CURRENT 20030113 and OpenPKG v1.2 and now
  -    keep the data verbatim.
  -    <p>
  -    Further research proved that due to fact that only half a megabyte
  -    of sources are unpacked and the remaining sources already come in
  -    "gzip" format the use of "compress" format for the whole thing
  -    actually enlarged the "<tt>openpkg-*.src.sh</tt>" file. For this
  -    reason we dropped compression for source bootstrap entirely
  -    beginning with CURRENT 20030114 and OpenPKG v1.2 and now keep the
  -    data verbatim.
  -</faq>
  -
   <faq id="uid-security"
        title="What is the fuzz about those multiple user/group ids in OpenPKG?">
       Starting with OpenPKG 1.1 the bootstrapping package ("openpkg") uses
  @@ -630,6 +598,69 @@
               VER-REL</tt>", provided by the packages "<tt>gcc</tt>",
               "<tt>gcc32</tt>", "<tt>gcc34</tt>", etc.
        </ul>
  +</faq>
  +
  +<faq id="source-focus"
  +     title="Why does OpenPKG focus on source RPMs only?">
  +   OpenPKG by design focuses on source RPMs and the building and installing
  +   directly from them. Binary RPMs are just an intermediate and temporary
  +   result in this approach. From our perspective, they exist just temporarily
  +   on the target machine or on our FTP server because of bootstrapping and for
  +   emergency situations only. There are multiple reasons for this. The most
  +   important are:
  +   <p>
  +   <ul>
  +     <li>Stability: binary RPMs are inherently weak when it comes to run-time
  +         stability. The reason is that there are <i>always</i> differences
  +         between the build and install host -- sometimes more, sometimes less.
  +         But just the smallest difference (versions of vendor shared
  +         libraries, different kernel patch-levels, system configuration
  +         differences, etc) can lead to a broken application on the install
  +         host due to inherited assumptions from the build host.
  +         <p>
  +         Example: build host has higher configured maximum allowed
  +         size for shared memory segments (usually because runs also a
  +         RDBMS), building OSSP mm on it determines this high limit and
  +         has to hard-code this into the package, package is deployed
  +         to install host and breaks horrible because install host has
  +         default maximum allowed size for shared memory. Bang! And OSSP
  +         mm has no chance, some parameters of a system cannot be easily
  +         determined under run-time. Instead they have to be determined
  +         in a complex way under configuration/build-time.
  +     <p>
  +     <li>Flexibility: OpenPKG source RPMs often have a bunch of build
  +         "options" (can be queried with <tt>rpm -qpi</tt>) for allowing one to
  +         build a package in multiple different <i>variants</i>. For instance,
  +         our Apache package has 55 boolean options. This allows you
  +         (theoretically) to build 2^55 = 36,028.797,018.963,968 different
  +         binary RPMs out of a single source RPM. There is always one
  +         combination which fits your situation well. A binary package has no
  +         more possibilities, it just was built for a fixed combination of
  +         options. And we doubt that neither a simple "no options enabled" nor
  +         an "all options enabled" Apache binary module would be sufficient for
  +         our community.
  +     <p>
  +     <li>Security: even with package signing by the OpenPKG project, from a
  +         security paranoia point of few, one never can really just
  +         trust a package -- neither a source nor binary one. For real
  +         security, every piece of software has to be audited, or at
  +         least be auditable. Source packages make this possible, binary
  +         packages make this completely impossible except you are trying
  +         to perfect the art of disassembling object code and reviewing
  +         complex algorithms on assembly layer.
  +   </ul>
  +</faq>
  +
  +<faq id="noarch"
  +     title="Why aren't there any architecture independent OpenPKG binary RPMs?">
  +     As <a href="#source-focus">the previous answer</a> explained, 
  +     OpenPKG by design focuses on source RPMs and the building and
  +     installing directly from them. Binary RPMs are just an intermediate
  +     and temporary result in this approach. Binary RPMs make too
  +     much trouble and so we try to reduce their issues and not to
  +     add extra organisational complexity by the introduction of
  +     architecture independent packages. Hence we do not distinguish
  +     between architecture independent and dependent binary packages.
   </faq>
   
   </ol>
  @@ .
______________________________________________________________________
The OpenPKG Project                                    www.openpkg.org
CVS Repository Commit List                     [EMAIL PROTECTED]

Reply via email to