paulb@20 | 1 | See docs/index.html for the libxml2dom documentation.
|
paulb@35 | 2 |
|
paulb@35 | 3 | Issues
|
paulb@35 | 4 | ------
|
paulb@35 | 5 |
|
paulb@35 | 6 | Use of importNode seems to cause some kind of memory issue, probably related
|
paulb@41 | 7 | to nodes being shared across documents. This was observed in libxml2 2.6.0 but
|
paulb@41 | 8 | appears to be fixed in libxml2 2.6.16.
|
paulb@50 | 9 |
|
paulb@50 | 10 | Even compared to minidom, importNode is very slow (even the
|
paulb@50 | 11 | libxml2dom.macrolib implementation, too). A way is needed to get libxml2 to do
|
paulb@50 | 12 | the node copying itself.
|
paulb@50 | 13 |
|
paulb@50 | 14 | Experiments
|
paulb@50 | 15 | -----------
|
paulb@50 | 16 |
|
paulb@50 | 17 | The libxml2macro.py program, along with the libxml2dom.macrolib package
|
paulb@50 | 18 | provide support for writing DOM-style code which is then translated to
|
paulb@50 | 19 | libxml2-style code before being compiled to normal Python modules. This
|
paulb@50 | 20 | special translation should eliminate the need for high-level wrapper objects
|
paulb@50 | 21 | in many cases, since the actual compiled code will be manipulating the libxml2
|
paulb@50 | 22 | objects directly - the principal exception being that of attribute nodes.
|