You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by David James <ja...@gmail.com> on 2005/08/02 16:06:16 UTC
Call for testing on the "python-bindings-improvements" branch
Anyone noticed the "python-bindings-improvements" branch yet?
We need your help!
1) Please take a look at the branch and let me know whether "make
check-swig-py" runs on your machine.
2) Please look at the code changes and send me your feedback. Any and
all comments are much appreciated!
For a list of changes, see:
http://svn.haxx.se/dev/archive-2005-08/0004.shtml
To download the branch, checkout from the following repository:
http://svn.collab.net/repos/svn/branches/python-bindings-improvements/
Instructions for building and compiling the branch can be found at:
http://svn.collab.net/repos/svn/branches/python-bindings-improvements/subversion/bindings/swig/INSTALL
Thanks!
David
--
David James -- http://www.cs.toronto.edu/~james
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Python properties destroyed when container Pool is garbage collected
Posted by "troy d. straszheim" <tr...@resophonic.com>.
Cool. That fixes it here too. The "real python strings" make debugging
easier, as well. Thanks.
-t
David James wrote:
> On 8/4/05, troy d. straszheim <tr...@resophonic.com> wrote:
>
>>Here's a new version: on my machine, at the end, it shows that things
>>are garbled... rhmmm. I'm going to try this on a linux box as well.
>>
>>-t
>>
>>#!/usr/bin/env python
>>
>>import sys
>>import os
>>import getopt
>>import pprint
>>
>>
>>from svn import fs, core, repos
>>
>>os.system("rm -rf /tmp/temp_repo temp_checkout")
>>os.system("svnadmin create /tmp/temp_repo")
>>os.system("svn co file:///tmp/temp_repo ./temp_checkout")
>>os.chdir("./temp_checkout")
>>os.system("svn propset svn:externals \"short /external\" .")
>>os.system("svn commit -m first_rev_has_external")
>>os.system("svn propset svn:externals \"this_one /is_really_really_really_long_and_comes_out_completely_garbled\" .")
>>os.system("svn commit -m next_rev_has_empty_external")
>>
>>repos_ptr = repos.open("/tmp/temp_repo", core.Pool())
>>fsob = repos.fs(repos_ptr)
>>
>>young_root = fs.revision_root(fsob, 2, core.Pool())
>>old_root = fs.revision_root(fsob, 1, core.Pool())
>>
>>young_props = fs.node_proplist(young_root, "/", core.Pool())
>>old_props = fs.node_proplist(old_root, "/", core.Pool())
>>
>>oldprop = old_props["svn:externals"]
>>assert(young_props.has_key("svn:externals"))
>>youngprop = young_props["svn:externals"]
>>if old_props["svn:externals"] != young_props["svn:externals"]:
>> print "gah."
>>print len(young_props["svn:externals"])
>>
>>print "this external should be garbled: >>>%(external)s<<<" % { 'external' : youngprop }
>
>
> Wow! Thanks for the detailed test case, Troy! I've fixed this by
> turning SVN strings into real python strings, instead of pointers to
> "pool memory".
>
> Fixed in r15595.
>
> Cheers,
>
> David
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Python properties destroyed when container Pool is garbage collected
Posted by David James <ja...@gmail.com>.
On 8/4/05, troy d. straszheim <tr...@resophonic.com> wrote:
> Here's a new version: on my machine, at the end, it shows that things
> are garbled... rhmmm. I'm going to try this on a linux box as well.
>
> -t
>
> #!/usr/bin/env python
>
> import sys
> import os
> import getopt
> import pprint
>
> #os.putenv("PYTHONPATH", "/Users/troy/Projects/svn-install/lib/svn-python")
> #os.putenv("DYLD_LIBRARY_PATH", "/Users/troy/Projects/svn-install/lib")
>
> from svn import fs, core, repos
>
> os.system("rm -rf /tmp/temp_repo temp_checkout")
> os.system("svnadmin create /tmp/temp_repo")
> os.system("svn co file:///tmp/temp_repo ./temp_checkout")
> os.chdir("./temp_checkout")
> os.system("svn propset svn:externals \"short /external\" .")
> os.system("svn commit -m first_rev_has_external")
> os.system("svn propset svn:externals \"this_one /is_really_really_really_long_and_comes_out_completely_garbled\" .")
> os.system("svn commit -m next_rev_has_empty_external")
>
> repos_ptr = repos.open("/tmp/temp_repo", core.Pool())
> fsob = repos.fs(repos_ptr)
>
> young_root = fs.revision_root(fsob, 2, core.Pool())
> old_root = fs.revision_root(fsob, 1, core.Pool())
>
> young_props = fs.node_proplist(young_root, "/", core.Pool())
> old_props = fs.node_proplist(old_root, "/", core.Pool())
>
> oldprop = old_props["svn:externals"]
> assert(young_props.has_key("svn:externals"))
> youngprop = young_props["svn:externals"]
> if old_props["svn:externals"] != young_props["svn:externals"]:
> print "gah."
> print len(young_props["svn:externals"])
>
> print "this external should be garbled: >>>%(external)s<<<" % { 'external' : youngprop }
Wow! Thanks for the detailed test case, Troy! I've fixed this by
turning SVN strings into real python strings, instead of pointers to
"pool memory".
Fixed in r15595.
Cheers,
David
--
David James -- http://www.cs.toronto.edu/~james
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Call for testing on the "python-bindings-improvements" branch
Posted by David James <ja...@gmail.com>.
On 8/3/05, troy d. straszheim <tr...@resophonic.com> wrote:
> Cool. This stuff works for me on OSX 10.4, gcc4, python 2.4.1.
> svnlook.py works, the two tests pass.
Fantastic! I don't have access to an OS X box, so your feedback is
especially valuable.
Cheers,
David
--
David James -- http://www.cs.toronto.edu/~james
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Call for testing on the "python-bindings-improvements" branch
Posted by "troy d. straszheim" <tr...@resophonic.com>.
Cool. This stuff works for me on OSX 10.4, gcc4, python 2.4.1.
svnlook.py works, the two tests pass.
-t
David James wrote:
> Anyone noticed the "python-bindings-improvements" branch yet?
>
> We need your help!
> 1) Please take a look at the branch and let me know whether "make
> check-swig-py" runs on your machine.
> 2) Please look at the code changes and send me your feedback. Any and
> all comments are much appreciated!
>
> For a list of changes, see:
> http://svn.haxx.se/dev/archive-2005-08/0004.shtml
> To download the branch, checkout from the following repository:
> http://svn.collab.net/repos/svn/branches/python-bindings-improvements/
> Instructions for building and compiling the branch can be found at:
> http://svn.collab.net/repos/svn/branches/python-bindings-improvements/subversion/bindings/swig/INSTALL
>
> Thanks!
>
> David
>
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Call for testing on the "python-bindings-improvements" branch
Posted by Christopher Ness <ch...@nesser.org>.
On Wed, 2005-08-03 at 11:02 -0400, David James wrote:
> On 8/2/05, Christopher Ness <ch...@nesser.org> wrote:
> > Ok, here is my configure info and some version info:
> Chris, thanks a lot for sending me this information! It is very helpful!
>
> I've committed fixes in r15565 and r15566. More details below.
No problem.
I updated to r15566 and recompiled/tested.
Things ran just fine. The tests passed.
<snip>
> > /home/nesscg/sandbox/python-bindings-improvements/subversion/include/svn_types.h:431: Warning(451): Setting const char * member may leak memory.
> > /home/nesscg/sandbox/python-bindings-improvements/subversion/include/svn_types.h:432: Warning(451): Setting const char * member may leak memory.
> > /home/nesscg/sandbox/python-bindings-improvements/subversion/include/svn_types.h:433: Warning(451): Setting const char * member may leak memory.
> > /home/nesscg/sandbox/python-bindings-improvements/subversion/include/svn_types.h:434: Warning(451): Setting const char * member may leak memory.
> > /home/nesscg/sandbox/python-bindings-improvements/subversion/include/svn_string.h:85: Warning(451): Setting const char * member may leak memory.
>
> These warnings signal potential memory leaks, because SWIG and Python
> don't keep track of the memory during Python->C variable assignments.
> This bug is present on trunk as well. We haven't come up with a good
> solution to this yet. Ideas?
I don't know as my experience with python is minimal and SWIG is
non-existent. But if SWIG and Python don't keep track of the memory,
than there is only one thing left to do it, right?
Cheers,
Chris
--
Wireless Group
McMaster University
summer
11:38:26 up 12 days, 21:31, 1 user, load average: 0.04, 0.05, 0.07
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Call for testing on the "python-bindings-improvements" branch
Posted by David James <ja...@gmail.com>.
On 8/2/05, Christopher Ness <ch...@nesser.org> wrote:
> Ok, here is my configure info and some version info:
Chris, thanks a lot for sending me this information! It is very helpful!
I've committed fixes in r15565 and r15566. More details below.
> [...]
> [nesscg@jackd python-bindings-improvements]$ python
> Python 2.2.3 (#1, Oct 15 2003, 23:33:35)
> [GCC 3.3.1 20030930 (Red Hat Linux 3.3.1-6)] on linux2
> [...]
> checking swig version... 1.3.24
> [...]
> [nesscg@jackd python-bindings-improvements]$ svnversion .
> 15551
>
> make check-swig-py 2>&1
> cd /home/nesscg/sandbox/python-bindings-improvements/subversion/bindings/swig/python; \
> /usr/bin/python2 /home/nesscg/sandbox/python-bindings-improvements/subversion/bindings/swig/python/tests/run_all.py
> EE
> ======================================================================
> ERROR: test_compatibility_layer (pool.PoolTestCase)
> ----------------------------------------------------------------------
> Traceback (most recent call last):
> File
> "/home/nesscg/sandbox/python-bindings-improvements/subversion/bindings/swig/python/tests/pool.py", line 53, in test_compatibility_layer
> self.assertTrue(pool_ref())
> AttributeError: PoolTestCase instance has no attribute 'assertTrue'
>
> ======================================================================
> ERROR: test_pool (pool.PoolTestCase)
> ----------------------------------------------------------------------
> Traceback (most recent call last):
> File
> "/home/nesscg/sandbox/python-bindings-improvements/subversion/bindings/swig/python/tests/pool.py", line 23, in test_pool
> self.assertTrue(parent_pool_ref())
> AttributeError: PoolTestCase instance has no attribute 'assertTrue'
>
> ----------------------------------------------------------------------
> Ran 2 tests in 0.013s
>
> FAILED (errors=2)
> make: *** [check-swig-py] Error 1
Good catch! I've used functionality which is only available in newer
versions of Python. I've fixed this for Python 2.2 in r15565.
>
> > 2) Please look at the code changes and send me your feedback. Any and
> > all comments are much appreciated!
>
> These were brought up by the compiler for `make swig-py`.
Thanks for the report! I'll try to fix as many of the warnings as I
can. See below.
/home/nesscg/sandbox/python-bindings-improvements/subversion/bindings/swig/python/libsvn_swig_py/swigutil_py.c:
In function `svn_swig_py_pool_destroyed':
> /home/nesscg/sandbox/python-bindings-improvements/subversion/bindings/swig/python/libsvn_swig_py/swigutil_py.c:144: warning: passing arg 2 of `PyObject_SetAttrString' discards qualifiers from pointer target type
Good catch. I forgot to properly typecast a pointer. Fixed in r15566.
> /home/nesscg/sandbox/python-bindings-improvements/subversion/include/svn_types.h:431: Warning(451): Setting const char * member may leak memory.
> /home/nesscg/sandbox/python-bindings-improvements/subversion/include/svn_types.h:432: Warning(451): Setting const char * member may leak memory.
> /home/nesscg/sandbox/python-bindings-improvements/subversion/include/svn_types.h:433: Warning(451): Setting const char * member may leak memory.
> /home/nesscg/sandbox/python-bindings-improvements/subversion/include/svn_types.h:434: Warning(451): Setting const char * member may leak memory.
> /home/nesscg/sandbox/python-bindings-improvements/subversion/include/svn_string.h:85: Warning(451): Setting const char * member may leak memory.
These warnings signal potential memory leaks, because SWIG and Python
don't keep track of the memory during Python->C variable assignments.
This bug is present on trunk as well. We haven't come up with a good
solution to this yet. Ideas?
The other warnings look relatively harmless.
>/home/nesscg/sandbox/python-bindings-improvements/subversion/include/svn_props.h:305:
Warning(305): Bad constant value (ignored).
This is caused by SWIG having trouble with a Subversion header file.
It still works, though.
/home/nesscg/sandbox/python-bindings-improvements/subversion/bindings/swig/python/libsvn_swig_py/swigutil_py.c:43:
> /usr/local/share/swig/1.3.24/python/pyrun.swg: In function
> `PySwigObject_GetType':
> /usr/local/share/swig/1.3.24/python/pyrun.swg:179: warning:
> initialization discards qualifiers from pointer target type
> /usr/local/share/swig/1.3.24/python/pyrun.swg: In function
> `PySwigPacked_GetType':
> /usr/local/share/swig/1.3.24/python/pyrun.swg:331: warning:
> initialization discards qualifiers from pointer target type
This looks like a harmless warning in the internals of SWIG 1.3.24 --
it doesn't show up in SWIG 1.3.25.
Cheers,
David
--
David James -- http://www.cs.toronto.edu/~james
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org