The dataforward.py program does not work under Windows XP, Python 2.4.3, Twisted
2.4.0.

I copied the example inside t.py. Once started, no echo is done when typing from the
console. The program does not read stdin and does not forward it in the connection.
I had to hit Ctrl-C to close it and you can see that what was typed at the console
was not taken by the program.

If twisted.internet.stdio.StandardIO is not working on Win32 systems, the book should say so.
...
D: mp>HEAD / HTTP/1.0
==> / <==
HEAD: error reading `/': Is a directory
HEAD: cannot open `HTTP/1.0' for reading: No such file or directory

Figure 3-1 actually indicates an error, not a successful validation as the main text suggests. In fact, an error _is_ returned if the sample code is executed, for reasons discussed below.

The format of multipart/form-data data blocks is given by section 7.2.1 of RFC 1521.

The validate.py code does not define a (required) "boundary" parameter value in the enclosing Content-Type header field. For the purposes of further discussion, however, this value will be assumed to be the 20 character string derived from the the randomChars list generated in encodeForm.

The encodeForm function does not create a properly-formatted multipart/form-data data block.

Within a multipart/form-data data block, each body part must begin with an encapsulation boundary, defined as two hyphens, followed by the "boundary" parameter value from the enclosing Content-Type header field, followed by a CRLF pair. The final part must be trailed by a special encapsulation boundary. This special encapsulation boundary has an additional two hyphens inserted before the final CRLF pair. All encapsulation boundaries, except the first, must be preceeded by CRLF pairs.

The validate.py code's encapsulation boundaries, with the exception of the first and last, are comprised of a 2 CRLF pairs, followed by 3 hyphens, the boundary string, another 3 hyphens, and a CRLF pair.

The validate.py code's first encapsulation boundary is like its others, except it lacks the 2 preceeding CRLF pairs.

The validate.py code's last encapsulation boundary is like its others, except it lacks a trailing CRLF.

None of these encapsulation boundaries is correct.

Additionally, the validate.py code does not insert blank lines between the header and body content of body parts.

Finally, the W3C validator service won't properly validate XHTML files, unless the relevant body part includes a Content-Type header specifing application/xhtml+xml. Such a header is not inserted by the validate.py code.

Two issues:

util.quote is deprecated, so it will not work. Plus, it's not a great way to insert the data.

More seriously, runQuery will return an error saying that there are no results to fetch. runOperation is the correct function.

Corrected code:
def render(self, request):
title = request.args['title'][0]
body = request.args['body'][0]
query = "Insert into posts (title, body) values (%s, %s)"
self.db.runOperation(query, (title, body)).addCallback(self._saved, request).addErrback(self._saveFailed, request)
return server.NOT_DONE_YET

return defer.succeed(requestedInterface, avatar, logout)

should be

return defer.succeed((requestedInterface, avatar, logout))

it has been corrected in the downloadable example code, but the book is wrong.

"With key-based authentication, the server is given a copy of a user's private key."

Hmmm, that would make the private key "un-private" wouldn't it? This is the exact opposite of what the server gets. In reality the server needs a copy of the user's public key.