summaryrefslogtreecommitdiffstats
path: root/Lib/test/test_support.py
Commit message (Collapse)AuthorAgeFilesLines
* Give more useful information about a failing PyUnit-style test.Fred Drake2001-07-161-2/+11
|
* Add support for Windows using "mbcs" as the default Unicode encoding when ↵Mark Hammond2001-05-131-0/+4
| | | | dealing with the file system. As discussed on python-dev and in patch 410465.
* Get rid of the superstitious "~" in dict hashing's "i = (~hash) & mask".Tim Peters2001-05-131-0/+8
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The comment following used to say: /* We use ~hash instead of hash, as degenerate hash functions, such as for ints <sigh>, can have lots of leading zeros. It's not really a performance risk, but better safe than sorry. 12-Dec-00 tim: so ~hash produces lots of leading ones instead -- what's the gain? */ That is, there was never a good reason for doing it. And to the contrary, as explained on Python-Dev last December, it tended to make the *sum* (i + incr) & mask (which is the first table index examined in case of collison) the same "too often" across distinct hashes. Changing to the simpler "i = hash & mask" reduced the number of string-dict collisions (== # number of times we go around the lookup for-loop) from about 6 million to 5 million during a full run of the test suite (these are approximate because the test suite does some random stuff from run to run). The number of collisions in non-string dicts also decreased, but not as dramatically. Note that this may, for a given dict, change the order (wrt previous releases) of entries exposed by .keys(), .values() and .items(). A number of std tests suffered bogus failures as a result. For dicts keyed by small ints, or (less so) by characters, the order is much more likely to be in increasing order of key now; e.g., >>> d = {} >>> for i in range(10): ... d[i] = i ... >>> d {0: 0, 1: 1, 2: 2, 3: 3, 4: 4, 5: 5, 6: 6, 7: 7, 8: 8, 9: 9} >>> Unfortunately. people may latch on to that in small examples and draw a bogus conclusion. test_support.py Moved test_extcall's sortdict() into test_support, made it stronger, and imported sortdict into other std tests that needed it. test_unicode.py Excluced cp875 from the "roundtrip over range(128)" test, because cp875 doesn't have a well-defined inverse for unicode("?", "cp875"). See Python-Dev for excruciating details. Cookie.py Chaged various output functions to sort dicts before building strings from them. test_extcall Fiddled the expected-result file. This remains sensitive to native dict ordering, because, e.g., if there are multiple errors in a keyword-arg dict (and test_extcall sets up many cases like that), the specific error Python complains about first depends on native dict ordering.
* In Jython, `@' is not allowed in module names. Extend the TESTFN testBarry Warsaw2001-03-231-3/+7
| | | | to use "$test" when in Jython. Closes SF patch #403668.
* When the regression test is run in verbose mode, make the PyUNIT-basedFred Drake2001-03-231-1/+1
| | | | tests a little noisier, providing more progress information.
* Updated to latest PyUnit version (1.31 in PyUnit CVS); test_support.pySteve Purcell2001-03-221-10/+6
| | | | changed accordingly.
* Just import sys at the top instead of inside lots of functions.Fred Drake2001-03-211-3/+35
| | | | Add some helpers for supporting PyUNIT-based unit testing.
* Oops. A RISCOS patch I forgot to check in.Guido van Rossum2001-03-131-1/+7
|
* Whitespace normalization.Tim Peters2001-02-211-1/+0
|
* Add test for syntax error on "x = 1 + 1".Jeremy Hylton2001-02-191-0/+9
| | | | Move check_syntax() function into test_support.
* docstring typoSkip Montanaro2001-01-201-1/+1
|
* Change verify() function to raise TestFailed, not AssertionError.Guido van Rossum2001-01-191-2/+2
| | | | | | (I realize that I didn't really test this, because all the tests succeed, so verify() never raised an AssertionError -- but the test suite still succeeds, so I'm not too worried.)
* Use constructor form of "raise"; normalize <wink> docstrings.Tim Peters2001-01-191-9/+5
|
* Whitespace normalization. Leaving tokenize_tests.py alone for now.Tim Peters2001-01-181-10/+10
|
* This patch removes all uses of "assert" in the regression test suiteMarc-André Lemburg2001-01-171-0/+12
| | | | | | | and replaces them with a new API verify(). As a result the regression suite will also perform its tests in optimization mode. Written by Marc-Andre Lemburg. Copyright assigned to Guido van Rossum.
* Update the code to better reflect recommended style:Fred Drake2000-12-121-1/+1
| | | | | Use != instead of <> since <> is documented as "obsolescent". Use "is" and "is not" when comparing with None or type objects.
* Make reindent.py happy (convert everything to 4-space indents!).Fred Drake2000-10-231-47/+47
|
* Restore TestSkipped exception; appears to have disappeared in last checkin.Fred Drake2000-07-241-3/+20
| | | | | Make both TextFailed and TestSkipped subclasses of Error, which derives from Exception. Docstrings have been added for the exceptions and module.
* make TestFailed a class exceptionSkip Montanaro2000-07-191-1/+2
|
* Changed runs of 8 spaces to tab -- to satisfy the tab nanny.Guido van Rossum1998-06-091-7/+7
|
* Move unified findfile() into test_support.pyGuido van Rossum1998-04-231-0/+12
|
* Add verbose flag for regression test to clear.Guido van Rossum1996-12-201-0/+2
|
* * os.py: _exit doesn't exist in all variations of posixGuido van Rossum1993-01-261-0/+19
| | | | * Added fcmp() to test_support.py and use it in test*.py
* Initial revisionGuido van Rossum1992-01-271-0/+22
generated by cgit v0.12 at 2025-09-19 13:38:06 (GMT)