[Lldb-commits] [PATCH] D43532: Fix TestSBData.py on Windows (which uses Python 3)
Adrian McCarthy via Phabricator via lldb-commits
lldb-commits at lists.llvm.org
Tue Feb 20 15:18:59 PST 2018
amccarth created this revision.
amccarth added a reviewer: zturner.
Herald added a subscriber: sanjoy.
This test uses the SB API to set and read back bytes of data, and it works fine when Python 2 is the scripting language. On Windows, however, Python 3 is the default.
Note this line from the test:
addr_data = '\x11\x22\x33\x44\x55\x66\x77\x88'
The intent here is to create an array of eight bytes as a string literal. This works in Python 2, but Python 3 treats those as a series Unicode code points, and the CPython implementation happens to encode those code points as UTF-8. The first seven characters encode directly as single bytes of the same value. But the UTF-8 encoding of 0x88 is two bytes long: 0xC2 0x88. So the input array doesn't have the intended data at the end, so the test cases that use all eight bytes fail.
Adding the `b` prefix tells Python 3 that we want a byte array rather than a string. With this change, the test now passes on Windows. I believe this also works for Python 2 (Python 2.7 accepts the b-prefix and seems to do the right thing), but I'm not very familiar with the details of Python 2.
@@ -25,7 +25,7 @@
"""Test the SBData::SetData() to ensure the byte order and address
byte size are obeyed"""
- addr_data = '\x11\x22\x33\x44\x55\x66\x77\x88'
+ addr_data = b'\x11\x22\x33\x44\x55\x66\x77\x88'
error = lldb.SBError()
data = lldb.SBData()
data.SetData(error, addr_data, lldb.eByteOrderBig, 4)
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 707 bytes
Desc: not available
More information about the lldb-commits