2.6 Merge Notes

  • Joseph Armbruster

    I would like to open up this thread to store some general notes on the merge I have ventured on with the trunk.  Specifically, the merge being performed is with:

    http://svn.python.org/projects/python/trunk @ 61395

    • Christopher Fairbairn

      Once the general merge is building etc it might also pay to include the "smartphone.patch" file available in the patch download from http://sourceforge.net/project/showfiles.php?group_id=104228&package_id=247631 (from the October 2007 release)

      This adds support for:
      1) optionally building with Visual Studio 2005 (or 2008)
      2) Windows Mobile Smartphone (i.e. non touchscreen devices)
      3) Windows Mobile 6.0 devices

      As well as some minor bug fixes and a slightly improved menu layout on Windows Mobile 5.0 and above devices.

      If this patch gets applied I also have a number of small additional patches which I never submitted that build on top of this to further improve smartphone support, and increase the usability of the Python console.

    • Joseph Armbruster

      I am currently using:
      scons-0.97.win32.exe and
      python-2.6a1.msi  which is rev 61155

      If you receive an error that states that the 'win32' platform was not found, do not fret!  As has been circulating on pydev, there are issues surrounding the msvcr and accompanying manifest.  This comes to light when attempting to build pythonce using scons.

      c:\python26\dlls contains a manifest:  Microsoft.VC90.CRT.manifest

      I needed to change the msvcr portion to:  <file name="../msvcr90.dll" />

      Unfortunately, scons does not propagate this error.

    • Joseph Armbruster

      In the unit test suite, at a first pass, these failed to compile.  The "bad" ones are suppose to fail, the others I have yet to look into... I'm pushing forward to test the new cab out now.  woot :-)

      - Joseph Armbruster



    • Joseph Armbruster

      Word of advice.  If you installed the python 2.6a1 off of python.org the magic numbers in import.c will be mis-matched.  Since your local version of python is what is compiling your .py  files into .pyc's, this may go unnoticed.... Until you try to import os, or something else and you get a ZipImportError. Unfortunately, this error does not tell you anything about magic numbers.  After digging a bit deeper, I found out that was the root cause.

      So, I hacked up my import.c for the time being to make it work.  I understand this is in fact BAD (and it will not be committed like this), however it was a quick fix to be able to use the latest 2.6a1 installer without fudging my tree :-)

      //#define MAGIC (62161 | ((long)'\r'<<16) | ((long)'\n'<<24))
      // Joe - for temporary compatibility with my installed version :-)
      #define MAGIC (62151 | ((long)'\r'<<16) | ((long)'\n'<<24))

    • kunicki

      kunicki - 2008-06-14

      Out of curiosity, has there been any progress on this update? It would be great to see this progress. Also out of curiosity, any thoughts about brining iPython to PPC?


Get latest updates about Open Source Projects, Conferences and News.

Sign up for the SourceForge newsletter:

No, thanks