summaryrefslogtreecommitdiff
path: root/doc/source/cdef.rst
diff options
context:
space:
mode:
authorArmin Rigo <arigo@tunes.org>2017-08-12 15:22:53 +0200
committerArmin Rigo <arigo@tunes.org>2017-08-12 15:22:53 +0200
commit5fd6db1f71f2b2eeb506e11dde2c288d7ff2cfd7 (patch)
treea339ea87442a9e728ed05139bc0dad445a7d0b38 /doc/source/cdef.rst
parente4a19d5af05071ced6def13342d50e5af112dd42 (diff)
downloadcffi-5fd6db1f71f2b2eeb506e11dde2c288d7ff2cfd7.tar.gz
Another note about performance
Diffstat (limited to 'doc/source/cdef.rst')
-rw-r--r--doc/source/cdef.rst4
1 files changed, 3 insertions, 1 deletions
diff --git a/doc/source/cdef.rst b/doc/source/cdef.rst
index 72d2edc..4d2f0c0 100644
--- a/doc/source/cdef.rst
+++ b/doc/source/cdef.rst
@@ -314,7 +314,9 @@ that ``libpath`` `cannot be None`__ on Windows with Python 3.
Let me state it again: this gives ABI-level access to the library, so
you need to have all types declared manually exactly as they were
while the library was made. No checking is done. Mismatches can
-cause random crashes.
+cause random crashes. API-level access, on the other hand, is safer.
+Speed-wise, API-level access is much faster (it is common to have
+the opposite misconception about performance).
Note that only functions and global variables live in library objects;
the types exist in the ``ffi`` instance independently of library objects.