Andreas Jung wrote:

--On 5. Januar 2006 12:47:20 +0100 Andreas Jung <[EMAIL PROTECTED]> wrote:

As you might know I worked on the integration of the Zope 3 ZPT
implementation for Zope 2.10. Before commiting the changes to the trunk I
would like discuss my approach for Zope 2.10

I forgot to mention a major point: compatibility.

When a ZPT is internally stored a unicode string then content returned by methods called through the ZPT will be implicitly converted to unicode. This will definitely raise UnicodeDecodeErrors. So how to deal with this issue?

Ah, I wrote my reply before reading this.

- allowing only unicode textual content when calling macros, PyScript etc.

- converting non-unicode to unicode inside the TAL code using some
  encoding. The encoding could be specified as property of the called
  method (function properties) or object.

In effect Python already does this, it just decodes to unicode using a strict ASCII encoding. Making this configurable per page template might be good, though I'm worried about supporting implicit behavior leading to bad coding patterns. I'd prefer code to be Python unicode clean, but allowing in, say, UTF-8 strings, into a page template and then implicitly converting them to unicode, is inviting people to persist in not understanding the way to write good unicode code.

We really _need_ to discuss this issue early to minimize side effects
and to be able to provide the best compatibility possible.




Zope-Dev maillist  -
**  No cross posts or HTML encoding!  **
(Related lists - )

Reply via email to