Commit Graph

63 Commits

Author SHA1 Message Date
Armin Rigo f879024487 test and fix for buggy handling of exceptions raised by C functions,
causing the profiler to crash on an AssertionError if the same Python
function catches multiple exceptions from C functions.
2005-09-20 18:50:13 +00:00
Martin v. Löwis a4dac4094a Patch #645894: Use getrusage for computing the time consumption in
profile.py if available.
2005-03-03 11:39:45 +00:00
Tim Peters b497c106d5 Whitespace normalization. 2005-01-10 16:48:37 +00:00
Johannes Gijsbers c0b194a770 Bug #489256: remove out of date and out of place profile.doc, and let
profile.help() point at the library reference instead of profile.doc.
2005-01-10 09:07:22 +00:00
Johannes Gijsbers 2abe785fee Move code in __name__ == '__main__' block into main() function, rewrite code so
there's no need to subclass OptionParser.
2005-01-09 01:58:02 +00:00
Nicholas Bastin 12ac3e1f49 Using repr() generates entries that the current stats package can't
collate, so setting it back to the function name
2004-07-12 23:38:02 +00:00
Nicholas Bastin 2e39d80925 Fix SF Bug #989066 2004-07-12 13:44:26 +00:00
Tim Peters 4e0e1b6a54 Whitespace normalization. 2004-07-07 20:54:48 +00:00
Nicholas Bastin c69ebe8d50 Enable the profiling of C functions (builtins and extensions) 2004-03-24 21:57:10 +00:00
Guido van Rossum 48713e8798 Add runctx to __all__. 2004-03-23 19:19:21 +00:00
Nicholas Bastin 824b1b2da8 Added command line options for profile.py - one for stats output file
and one for sort order when using stdout.  Uses optparse.
2004-03-23 18:44:39 +00:00
Nicholas Bastin 1eb4bfc657 Added global runctx function to profile to fix SF Bug #716587 2004-03-22 20:12:56 +00:00
Walter Dörwald 70a6b49821 Replace backticks with repr() or "%r"
From SF patch #852334.
2004-02-12 17:35:32 +00:00
Guido van Rossum 38443c1513 Remove unneeded import. 2003-10-22 17:22:18 +00:00
Raymond Hettinger 97aa32b467 Replace a reduce() with sum(). 2003-10-22 16:49:01 +00:00
Guido van Rossum 68468eba63 Get rid of many apply() calls. 2003-02-27 20:14:51 +00:00
Raymond Hettinger e0d4972acc Replaced .keys() with dictionary iterators 2002-06-02 18:55:56 +00:00
Raymond Hettinger 16e3c427f3 Replace boolean test with is None. 2002-06-01 16:07:16 +00:00
Raymond Hettinger 54f0222547 SF 563203. Replaced 'has_key()' with 'in'. 2002-06-01 14:18:47 +00:00
Fred Drake ee836445d1 Added a missing period at the end of an error message. 2001-12-05 22:27:47 +00:00
Fred Drake d10ed8b179 Minor code cleanups based on comments from Neal Norwitz. 2001-10-17 01:49:50 +00:00
Tim Peters e13cc92e6c Repair key stutter + auto-complete ugliness. 2001-10-09 21:01:31 +00:00
Tim Peters 659a60311d Allow the profiler's calibration constant to be specified in the constructor
call, or via setting an instance or class vrbl.
Rewrote the calibration docs.
Modern boxes are so friggin' fast, and a profiler event does so much work
anyway, that the cost of looking up an instance vrbl (the bias constant)
per profile event just isn't a big deal.
2001-10-09 20:51:19 +00:00
Tim Peters cce092d046 A brand new implementation of Profile.calibrate(). This measures an
actual run of the profiler, instead of timing a simplified simulation of
part of what the profiler does.  It computes a constant about 60% higher
on my Win98SE box than the old method, and the new constant appears much
more realistic.  Deleted the undocumented simple(), instrumented(), and
profiler_simulation() methods (which existed only to support the previous
calibration method).
2001-10-09 05:31:56 +00:00
Tim Peters fb163784ab Typo repair in comment. 2001-10-07 08:49:02 +00:00
Tim Peters 8d061ed75b Guido points out that the comments for self.cur[2] were subtly but
seriously wrong.  This started out by just fixing the docs, but then it
occurred to me that the doc confusion propagated into misleading vrbl names
too, so I also renamed those to match reality.  As a result, INO the time
computations are much easier to understand now (within the limitations of
vast quantities of 3-character names <wink>).
2001-10-07 08:35:44 +00:00
Tim Peters db1ed2aec3 At Guido's request, changed the code that's conceptually asserting stuff
to use assert stmts (was raising unexpected kinds of exceptions).
2001-10-07 04:30:53 +00:00
Tim Peters 6e22149cb6 Repair some longstanding comment errors:
+ The last index in the timing tuple is 4, not 5 (noted by Guido).

+ The poorly named trace_dispatch_i works with float return values too.
2001-10-07 04:02:36 +00:00
Tim Peters 0a1fc4e389 Remove code and docs for the OldProfile and HotProfile classes: code
hasn't worked in years, docs were wrong, and they aren't interesting
anymore regardless.
2001-10-07 03:12:08 +00:00
Tim Peters df5cfd884d The fix to profile semantics broke the miserable but advertised way to
derive Profile subclasses.  This patch repairs that, restoring
negative tuple indices.  Yuck?  You bet.
2001-10-05 23:15:10 +00:00
Guido van Rossum f137f75ab8 Hopefully fix the profiler right. Add a test suite that checks that
it deals correctly with some anomalous cases; according to this test
suite I've fixed it right.

The anomalous cases had to do with 'exception' events: these aren't
generated when they would be most helpful, and the profiler has to
work hard to recover the right information.  The problems occur when C
code (such as hasattr(), which is used as the example here) calls back
into Python code and clears an exception raised by that Python code.
Consider this example:

    def foo():
        hasattr(obj, "bar")

Where obj is an instance from a class like this:

    class C:
        def __getattr__(self, name):
            raise AttributeError

The profiler sees the following sequence of events:

    call (foo)
    call (__getattr__)
    exception (in __getattr__)
    return (from foo)

Previously, the profiler would assume the return event returned from
__getattr__. An if statement checking for this condition and raising
an exception was commented out...  This version does the right thing.
2001-10-04 00:58:24 +00:00
Fred Drake a0bc9993e7 Undo previous patch; it did not quite work out. 2001-10-03 21:12:32 +00:00
Fred Drake 5a28bfbbc7 Change the sense of a test in how the profiler interprets exception events.
This should fix a bug in how time is allocated during exception propogation
(esp. in the presence of finally clauses).
2001-09-27 16:28:42 +00:00
Guido van Rossum e4deb959cc Fix two bugs detected by PyChecker: there's no need for redundant
"import MacOS", and there *is* a need for "import operator".
2001-08-09 21:22:15 +00:00
Jack Jansen 1bdcadd610 An import MacOS was missing after the code-rearranging. Added. 2001-06-19 20:11:36 +00:00
Fred Drake edb5ffb2c1 Performance improvements to the profiler:
Ensure that all the default timers are called as functions, not an
expensive method wrapper around a variety of different functions.

Agressively avoid dictionary lookups.

Modify the dispatch scheme (Profile.trace_dispatch_*(), where * is not
'call', 'exception' or 'return') so that the callables dispatched to
are simple functions and not bound methods -- this reduces the number
of layers of Python call machinery that gets touched.

Remove a couple of duplicate imports from the "if __name__ == ..."
section.

This closes SF patch #430948.
2001-06-08 04:25:24 +00:00
Jeremy Hylton adcf8a05a4 Add doc string for run from profile.doc. (pydoc motivates me to write
good doc strings.)

Fix silly argument handling; was using *args but really wanted 1
optional arg.

XXX Should profile.doc be merged into the documentation and removed
from the Lib directory?
2001-03-14 20:01:19 +00:00
Skip Montanaro c62c81e013 __all__ for several more modules 2001-02-12 02:00:42 +00:00
Tim Peters 2344fae6d0 Whitespace normalization. 2001-01-15 00:50:52 +00:00
Peter Schneider-Kamp 10e1bf2f64 remove all occurence of math.rint() from the sources
(and yes, "Currintly" also counts <0.5 wink>)
2000-08-10 04:23:30 +00:00
Thomas Wouters 7e47402264 Spelling fixes supplied by Rob W. W. Hooft. All these are fixes in either
comments, docstrings or error messages. I fixed two minor things in
test_winreg.py ("didn't" -> "Didn't" and "Didnt" -> "Didn't").

There is a minor style issue involved: Guido seems to have preferred English
grammar (behaviour, honour) in a couple places. This patch changes that to
American, which is the more prominent style in the source. I prefer English
myself, so if English is preferred, I'd be happy to supply a patch myself ;)
2000-07-16 12:04:32 +00:00
Guido van Rossum 54f22ed30b More trivial comment -> docstring transformations by Ka-Ping Yee,
who writes:

Here is batch 2, as a big collection of CVS context diffs.
Along with moving comments into docstrings, i've added a
couple of missing docstrings and attempted to make sure more
module docstrings begin with a one-line summary.

I did not add docstrings to the methods in profile.py for
fear of upsetting any careful optimizations there, though
i did move class documentation into class docstrings.

The convention i'm using is to leave credits/version/copyright
type of stuff in # comments, and move the rest of the descriptive
stuff about module usage into module docstrings.  Hope this is
okay.
2000-02-04 15:10:34 +00:00
Guido van Rossum 138bdafa8d Typo. (Andrew Dalke, without kjpylint) 1999-05-03 18:13:29 +00:00
Guido van Rossum 6952f17f31 No need to import string. (Andrew Dalke & kjpylint) 1999-05-03 18:13:00 +00:00
Guido van Rossum 4ecd85aad7 After the previous changes, func_normalize() turned out to be redundant.
This simplified some other places in the code.
1998-09-21 17:40:47 +00:00
Guido van Rossum b0a94c0549 Changes so that JPython can also use this version of profile.py.
Suggested by Jim Hugunin.
1998-09-21 16:52:44 +00:00
Guido van Rossum e3f8a64906 Comment out 't = t[0] + t[1]' in profiler_simulation() -- this
function is only used when running the calibration code, and it turns
out that recent changes in the timing code caused this statement to
raise an exception.
1998-09-21 14:52:22 +00:00
Guido van Rossum 8ca842066c A few lines were indented using spaces instead of tabs -- fix them. 1998-03-26 20:56:10 +00:00
Guido van Rossum 96c07fefa9 Prefer clock() over times() for timer function, except on the Mac,
where we use GetTicks() -- its clock() is a crock, with only 1 second
accuracy, I believe.
1998-03-17 14:46:43 +00:00
Guido van Rossum cbf3dd53b4 Use better timer on the mac
Open files in binary mode
(Jack)
1997-10-08 15:23:02 +00:00