Doesn't work dereference of sha1 to object.

[AlexSokol]

Then I try to use the program from here https://bitbucket.org/eugeneai/linux-project.git, Python returns this message:

Traceback (most recent call last):
File "/home/alex/convert.py", line 22, in
print i, comm.message.strip(), comm.committer
File "/usr/local/lib/python2.7/dist-packages/gitdb/util.py", line 238, in __ getattr __
self.set_cache(attr)

[Byron]

Unfortunately this is only a partial stack trace, hence it does not help.
gitdb has been phased out as well, and merged into gitpython.

[AlexSokol]

Ok, that is full stack trace:

Traceback (most recent call last):
File "/home/alex/convert.py", line 22, in
print i, comm.message.strip(), comm.committer
File "/usr/local/lib/python2.7/dist-packages/gitdb/util.py", line 238, in
getattr
self.set_cache(attr)
File
"/usr/local/lib/python2.7/dist-packages/GitPython-0.3.2.RC1-py2.7.egg/git/objects/commit.py",
line 131, in set_cache
binsha, typename, self.size, stream = self.repo.odb.stream(self.binsha)
File "/usr/local/lib/python2.7/dist-packages/gitdb/db/base.py", line 259,
in stream
return self._db_query(sha).stream(sha)
File "/usr/local/lib/python2.7/dist-packages/gitdb/db/base.py", line 239,
in _db_query
if db.has_object(sha):
File "/usr/local/lib/python2.7/dist-packages/gitdb/db/pack.py", line 88,
in has_object
self._pack_info(sha)
File "/usr/local/lib/python2.7/dist-packages/gitdb/db/pack.py", line 71,
in _pack_info
index = item2
File "/usr/local/lib/python2.7/dist-packages/gitdb/pack.py", line 419, in
sha_to_index
get_sha = self.sha
File "/usr/local/lib/python2.7/dist-packages/gitdb/util.py", line 238, in
getattr
self.set_cache(attr)
File "/usr/local/lib/python2.7/dist-packages/gitdb/pack.py", line 281, in
set_cache
mmap = self._cursor.map()
File "/usr/local/lib/python2.7/dist-packages/gitdb/util.py", line 238, in
getattr
self.set_cache(attr)
File "/usr/local/lib/python2.7/dist-packages/gitdb/pack.py", line 274, in
set_cache
raise AssertionError("The index file at %s is too large to fit into a
mapped window (%i > %i). This is a limitation of the implementation" %
(self._indexpath, self._cursor.file_size(), mman.window_size()))
AssertionError: The index file at
/home/alex/linux/.git/objects/pack/pack-5e55a3f93ec5edf76bbe7206d6fc3814e13c5b58.idx
is too large to fit into a mapped window (65351336 > 33554432). This is a
limitation of the implementation

I use GitPython-0.3.2.RC1, Python 2.7 and Ubuntu 12.04.
And I can take the list of commits (headcommit.hexsha), but then I try to
use headcommit.parents or something else this problem arises.

[Byron]

You have a rather big index (~65MB), but use gitpython on a 32 bit system. This means it will map only 32 MB chunks at once into memory. On a 64 bit system, it would use up to 1024 MB per mapping.
The issue you ran into is a sanity check as the implementation makes assumptions to make things easier on the implementors side.
The best workaround would be to use a 64 bit system or adjust the code to set the window size to something more suitable for your index file.

[AlexSokol]

Thanks for your answer.

[AlexSokol]

When I use 64 bit sistem, my programm works up to 15th commit, but then

Traceback (most recent call last):
File "/home/alex/convert.py", line 17, in
b = comm.parents
File "/usr/local/lib/python2.7/dist-packages/gitdb/util.py", line 238, in getattr
self.set_cache(attr)
File "/usr/local/lib/python2.7/dist-packages/GitPython-0.3.2.RC1-py2.7.egg/git/objects/commit.py", line 132, in >set_cache
self._deserialize(StringIO(stream.read()))
File "/usr/local/lib/python2.7/dist-packages/GitPython-0.3.2.RC1-py2.7.egg/git/objects/commit.py", line 443, in >_deserialize
self.author.name = self.author.name.decode(self.encoding)
LookupError: unknown encoding: object 320cfa6ce0b3dc794fedfa4bae54c0f65077234d

And what does "window size" mean and how it can be edited?

[Byron]

It seems that it parsed the commit's encoding incorrectly, as it has a hash string in a place where something like UTF-8 would be expected. Therefore the author's name cannot be encoded.
Apparently gitdb fails to parse the commit object correctly.

As a general workaround, you can just use a different Repository backend. It defaults to gitdb, but you may use GitCmd as well. Please have a look at the git.repo module on how to do that.