summaryrefslogtreecommitdiff
path: root/tests/data/utftest_utf16_le_nodecl.xml
diff options
context:
space:
mode:
authorArseny Kapoulkine <arseny.kapoulkine@gmail.com>2014-02-10 16:57:04 +0000
committerArseny Kapoulkine <arseny.kapoulkine@gmail.com>2014-02-10 16:57:04 +0000
commit79fb68ac4177206e063f8f29113abbe82ac49698 (patch)
treec9f07c8d3b6bc82c944c8c16c72f50bf374e7b6d /tests/data/utftest_utf16_le_nodecl.xml
parent9ba26b94c74a03ac937a5d5972f8f12a2916f301 (diff)
Use a null-terminated buffer for parsing as often as possible.
Parsing used to work on a non null-terminated buffer, inserting a fake null terminator to increase performance. This makes it impossible to implement fragment parsing that preserves PCDATA contents (as witnessed by some tests for boundary conditions that actually depended on this behavior). Since almost all uses result in us allocating an internal buffer anyway, the new policy is to make sure all buffers that are allocated by pugixml are null-terminated - the only exception now is external calls to load_buffer_inplace that don't trigger encoding conversion. git-svn-id: https://pugixml.googlecode.com/svn/trunk@977 99668b35-9821-0410-8761-19e4c4f06640
Diffstat (limited to 'tests/data/utftest_utf16_le_nodecl.xml')
0 files changed, 0 insertions, 0 deletions