[Python-checkins] python/dist/src README,1.164,1.165

Fri, 03 Jan 2003 20:05:53 -0800

Update of /cvsroot/python/python/dist/src
In directory sc8-pr-cvs1:/tmp/cvs-serv19177

Modified Files:
	README 
Log Message:
Merged Misc/AtheOS-NOTES into the platform-specific section.  Rewrote the
bsddb module build note to reflect the inclusion of bsddb3 and the demotion
of the old bsddb module.

Index: README
===================================================================
RCS file: /cvsroot/python/python/dist/src/README,v
retrieving revision 1.164
retrieving revision 1.165
diff -C2 -d -r1.164 -r1.165
*** README	3 Jan 2003 16:26:23 -0000	1.164
--- README	4 Jan 2003 04:05:51 -0000	1.165
***************
*** 595,640 ****
          versions would be appreciated!

! Configuring the bsddb and dbm modules
! -------------------------------------

! XXX Shouldn't this section be rewritten now that we use Sleepycat's
! BSDDB 4.0?

! Configuring the bsddb module can sometimes be a bit tricky.  This module
! provides a Python interface to the Berkeley DB library.  As of this writing
! several versions of the underlying library are in common use (versions 1.85,
! 2.x, 3.x, and 4.x).  The file formats across the various versions tend to be
! incompatible.  Some Linux distributions install multiple versions by
! default.  It is important that compatible versions of header files and
! libraries are used when building bsddb.  To make matters worse, version 1.85
! of Berkeley DB has known bugs in its hash file implementation, but is still
! the most widely available version of the library.  Many people build bsddb
! with version 1.85 but aren't aware of the bugs.  This affects people using
! the anydbm and dbhash modules because they are both use Berkeley DB's hash
! file format as a side effect of calling bsddb.hashopen.

! To try and remedy this problem, beginning with Python version 2.3 a number
! of changes to the bsddb build process were made.  First, and most important,
! the bsddb module will not be built with version 1.85 unless the relevant
! lines in setup.py are uncommented first and no other higher-numbered
! versions are found.  Second, matching versions of the library and include
! files must be found.  Third, searching is performed in order, starting from
! version 4 and proceeding to version 2 (or version 1 if it is enabled).
! Version-independent libraries and header files (e.g. /usr/lib/libdb.a and
! /usr/include/db.h) are never considered.  They must be in version-specific
! directories or have version-specific filenames (e.g. /usr/lib/libdb-3.2.so
! and /usr/include/db3/db_185.h).

! Since the bsddb module is programmed using the Berkeley DB version 1 API,
! the underlying library must be configured with the --enable-compat185 flag.
! Most vendor-provided distributions are so-configured.  This is generally
! only an issue if you build Berkeley DB from source.

- All this affects the dbm module as well.  There are several dbm-compliant
- APIs provided by different libraries, including ndbm, gdbm and Berkeley DB.
- The build process for dbm would previously use the version 1.85 library,
- thus extending the potential hash file corruption to the dbm module as well.
- The dbm module will use the library and include files found for the bsddb
- module if neither ndbm nor gdbm libraries are found.

  Configuring threads
--- 595,680 ----
          versions would be appreciated!

! AtheOS: From Octavian Cerna <tavy at ylabs.com>:

! 	Before building:

! 	    Make sure you have shared versions of the libraries you
! 	    want to use with Python. You will have to compile them
! 	    yourself, or download precompiled packages.

! 	    Recommended libraries:

! 		ncurses-4.2
! 		readline-4.2a
! 		zlib-1.1.4
! 
! 	Build:
! 
! 	    $ ./configure --prefix=/usr/python
! 	    $ make
! 
! 	    Python is always built as a shared library, otherwise
! 	    dynamic loading would not work.
! 
! 	Testing:
! 
! 	    $ make test
! 
! 	Install:
! 
! 	    # make install
! 	    # pkgmanager -a /usr/python
! 
! 
! 	AtheOS issues:
! 
! 	    - large file support: due to a stdio bug in glibc/libio,
! 	      access to large files may not work correctly.  fseeko()
! 	      tries to seek to a negative offset.  ftello() returns a
! 	      negative offset, it looks like a 32->64bit
! 	      sign-extension issue.  The lowlevel functions (open,
! 	      lseek, etc) are OK.
! 	    - sockets: AF_UNIX is defined in the C library and in
! 	      Python, but not implemented in the system.
! 	    - select: poll is available in the C library, but does not
! 	      work (It does not return POLLNVAL for bad fds and
! 	      hangs).
! 	    - posix: statvfs and fstatvfs always return ENOSYS.
! 	    - disabled modules:
! 		- mmap: not yet implemented in AtheOS
! 		- nis: broken (on an unconfigured system
! 		  yp_get_default_domain() returns junk instead of
! 		  error)
! 		- dl: dynamic loading doesn't work via dlopen()
! 		- resource: getrimit and setrlimit are not yet
! 		  implemented
! 
! 	    - if you are getting segmentation faults, you probably are
! 	      low on memory.  AtheOS doesn't handle very well an
! 	      out-of-memory condition and simply SEGVs the process.
! 
! 	Tested on:
! 
! 	    AtheOS-0.3.7
! 	    gcc-2.95
! 	    binutils-2.10
! 	    make-3.78
! 
! 
! Configuring the bsddb and dbm modules
! -------------------------------------
! 
! Beginning with Python version 2.3, the PyBsddb package
! <http://pybsddb.sf.net/> was adopted into Python as the bsddb package,
! exposing a set of package-level functions which provide
! backwards-compatible behavior.  Only versions 3.1 through 4.1 of
! Sleepycat's libraries provide the necessary API, so older versions
! aren't supported through this interface.  The old bsddb module has
! been retained as bsddb185, though it is not built by default.  Users
! wishing to use it will have to tweak Modules/Setup to build it.  The
! dbm module will still be built against the Sleepycat libraries if
! other preferred alternatives (ndbm, gdbm) are not found, though
! versions of the Sleepycat library prior to 3.1 are not considered.

  Configuring threads