[[[ To any NSA and FBI agents reading my email: please consider ]]] [[[ whether defending the US Constitution against all enemies, ]]] [[[ foreign or domestic, requires you to follow Snowden's example. ]]]
> AJAX returning json as application/json or the like will not get > analyzed by the HTML parser, and so it won't break. However many people > will always return the wrong mimetype, and in order to have LibreJS > function optimally, I think we will need to improve the "content > sniffer" part of LibreJS to determine what a response is _really_, > regardless of the content-type name. Maybe so, but here is a crucial point: DO NOT treat AJAX responses the same as pages! An AJAX request response might be JSON. A page won't be. Thus, the check "is this really JSON" should be done on AJAX responses but not on pages. > Another technique would be to > analyze the content of the response, make modifications to the document > structure if needed, but instead of serializing the result back, to use > a search/replace or diff algorithm to use the original response and > update it. That sounds like a good plan, but since it will take more work, I suggest implementing the first solution right away and putting the second on the TODO list. -- Dr Richard Stallman President, Free Software Foundation 51 Franklin St Boston MA 02110 USA www.fsf.org www.gnu.org Skype: No way! That's nonfree (freedom-denying) software. Use Ekiga or an ordinary phone call.
