You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I don't think it is safe to remove all *.pickle files in the source
directory; even if there aren't any other useful .pickle files right now,
there could be in the future (e.g. for pickle testing).
I'd prefer something more targeted.
Are these .pickle files generated during the build process, or just during
testing? Maybe it should be the responsibility of the 2to3 tests to clean
up after themselves?
drukker commented on Sep 24, 2009
drukkermannequin
MannequinAuthor
More actions
These .pickle files aren't created by the tests themselves, but they do
show up after running 'make test', or more specifically after running
'./python Lib/test/regrtest.py -v test_lib2to3'.
This is because a grammar generated from a .txt grammar file is cached
as a pickle for faster access by load_grammar in Lib/lib2to3/pgen2/driver.py. A pickle file then ends up in the same
directory as the .txt file.
IMHO, though it may be preferable to remove just the pickle files in Lib/lib2to3, it should never be required for tests to have pickle files
already available in the source tree. Instead, the test should create a
temporary pickle file, e.g. from a bytes object. Furthermore, binary
files shouldn't be versioned without good reason. Thus, removing all
pickle files should be safe for the future.
IMHO, though it may be preferable to remove just the pickle files in Lib/lib2to3,
Sounds reasonable.
it should never be required for tests to have pickle files
already available in the source tree.
Well, I don't know about 'should', but the current tests already do
make use of pickle files in the source tree: see Lib/test/*.pck.
So there are already versioned pickle files (but not .pickle files :-).
drukker commented on Sep 24, 2009
drukkermannequin
MannequinAuthor
More actions
You are right. Guess I was being a little too dogmatic. :-)
I hadn't found those .pck files, because I was only looking for binary
files.
Activity
drukker commentedon Sep 23, 2009
make clean and make distclean don't remove the grammar pickles generated
by load_grammar in Lib/lib2to3/pgen2/driver.py
(Lib/lib2to3/Grammar3.2.0.alpha.0.pickle,
Lib/lib2to3/PatternGrammar3.2.0.alpha.0.pickle).
Proposed patch attached. It removes all *.pickle files, which I assume
is safe. (Patch also corrects a comment.)
mdickinson commentedon Sep 24, 2009
I don't think it is safe to remove all *.pickle files in the source
directory; even if there aren't any other useful .pickle files right now,
there could be in the future (e.g. for pickle testing).
I'd prefer something more targeted.
Are these .pickle files generated during the build process, or just during
testing? Maybe it should be the responsibility of the 2to3 tests to clean
up after themselves?
drukker commentedon Sep 24, 2009
These .pickle files aren't created by the tests themselves, but they do
show up after running 'make test', or more specifically after running
'./python Lib/test/regrtest.py -v test_lib2to3'.
This is because a grammar generated from a .txt grammar file is cached
as a pickle for faster access by load_grammar in
Lib/lib2to3/pgen2/driver.py. A pickle file then ends up in the same
directory as the .txt file.
IMHO, though it may be preferable to remove just the pickle files in
Lib/lib2to3, it should never be required for tests to have pickle files
already available in the source tree. Instead, the test should create a
temporary pickle file, e.g. from a bytes object. Furthermore, binary
files shouldn't be versioned without good reason. Thus, removing all
pickle files should be safe for the future.
mdickinson commentedon Sep 24, 2009
Sounds reasonable.
Well, I don't know about 'should', but the current tests already do
make use of pickle files in the source tree: see Lib/test/*.pck.
So there are already versioned pickle files (but not .pickle files :-).
drukker commentedon Sep 24, 2009
You are right. Guess I was being a little too dogmatic. :-)
I hadn't found those .pck files, because I was only looking for binary
files.
Here's a new patch proposal.
mdickinson commentedon Sep 24, 2009
Thanks! I'll apply this later today, unless Benjamin gets there first. :)
mdickinson commentedon Sep 24, 2009
Committed in r75047 (trunk). Merged in r75048 (release26-maint), r75049
(py3k) and r75050 (release31-maint).