Hello,
last week I promised to send
out a description how to incorporate Pie-thon (a.k.a. parrotbench) and Pystone
tests into the IronPython distribution to run the tests along with the
IronPython test suite. Here is how to do it.
IRONPYTHON_BASE
├───External
│ ├───Parrotbench
│ └───Regress
│ └───Lib
└───Src
├───bin
├───Examples
├───IronMath
├───IronPython
│ ├───AST
│ ├───Hosting
│ ├───Modules
│ └───Objects
├───IronPythonConsole
├───IronPythonTest
└───Scripts
"External" directory contains all external tests:
- CPython library + tests inside the "Regress" subdirectory
- Parrotbench test in the "Parrotbench" subdirectory
"Src" is the IronPython code that we release.
- For the purpose of release we rename the directory (for example "IronPython-0.9"),
renaming it back to "Src" will plug it right in place.
"Parrotbench" directory contains the parrotbench
benchmark files (b*.py, parrotrun.py, ...)
To get parrotbench sources,
download IronPython 0.6 from http://www.ironpython.com.
The IronPython 0.6 distribution contains the Parrotbench in its
"tests" directory.
The parrotrun.py is a file Jim wrote to make running the tests easier. It has not been part of our releases yet and unfortunatelly I modified it so the version you get from IronPython 0.6 won't work. At the bottom of this email is the right version of the parrotrun.py that will work with the tests. Starting with our next release, we will include the parrotrun.py with the release so that things work smoother.
Pystone + CPython regression test
suite
Download the 2.3.5 release of CPython from http://www.python.org/2.3.5
Library and tests are in the "Lib" directory of the Python
installation.
Copy the whole Lib directory into "Regress".
Once you have combined IronPython release files with parrot-bench from IronPython 0.6 and standard CPython libraries, you can run the tests by exe
Running the tests:
To run the tests, execute:
Src\bin\IronPythonConsole.exe
Src\Scripts\RunTests.py
The RunTests.py accepts command line parameters, one of them being the type of the test - short, medium, long (run "RunTests.py -?" for more details)
Known issue is running "long" version of regression tests (runtests.py regr long). We had to make some changes to the Python test suite to get the long version running so if you run it, it will fail. For now, I would recommend running:
"IronPythonConsole.exe RunTests.py iron pystone parrot long"
I hope this helps for now and if you have any questions, don't hesitate to contact me so that we can address them.
Thanks
and keep in touch,
Martin
parrotrun.py file:
##################################################
"""
This provides a more convenient harness for running this
benchmark and collecting separate timings for each
component.
"""
from
time import clock
import sys
def
main(type="short"):
t0 = clock()
import b0
import b1
import
b2
import b3
import
b4
import b5
import b6
print 'import time = %.2f' %
(clock()-t0)
if type ==
"short":
N =
1
tests =
[b0,b1,b2,b3,b4,b6]
elif type ==
"medium":
N =
2
tests =
[b0,b1,b2,b3,b4,b5,b6]
elif type ==
"long":
N =
4
tests =
[b0,b1,b2,b3,b4,b5,b6]
else:
raise
AssertionError("Unknown test type: %s" % type)
results = {}
t0 = clock()
for i in range(N):
for test in
tests:
ts0
= clock()
test.main()
tm =
(clock()-ts0)
results.setdefault(test,
[]).append(tm)
print '%.2f sec running %s' % ( tm, test.__name__)
for test in
tests:
print '%s = %f -- %r' %
(test.__name__, sum(results[test])/N, results[test])
print 'all done in %.2f sec' % (clock()-t0)
if
__name__=="__main__":
kind = "short"
if len(sys.argv) > 1:
kind =
sys.argv[1]
main(kind)
##################################################
_______________________________________________ users-ironpython.com mailing list users-ironpython.com@lists.ironpython.com http://lists.ironpython.com/listinfo.cgi/users-ironpython.com
