ZCatalog regression: indexing None values
yuppie <[email protected]>
| Newsgroups | gmane.comp.web.zope.devel |
|---|---|
| Message-ID | <[email protected]> |
Hi! Indexes shipped with ZCatalog 3 no longer work with None values: https://github.com/zopefoundation/Products.ZCatalog/commit/c378cdab2fb8997af1a17261458a1528e7131243 I'm not sure if the old behavior did follow a well defined policy, I just found this explanation: http://stackoverflow.com/a/11224076 I think in general the policy was to ignore None values if the index can't handle them. And AFAICS that's a much better policy than raising errors as ZCatalog 3 does. There is nothing wrong about None values returned by content attributes. Often they have a special meaning like not initialized, not selected, not specified. In most cases it wouldn't cause trouble if None is not indexed, but it causes trouble if indexing raises errors. Adjusting code that implements special behavior for None, adjusting persistent attributes that contain None or adding wrappers that modify indexed attributes is a lot of work. I propose to ignore None values instead of raising errors. Cheers, Yuppie _______________________________________________ Zope-Dev maillist - [email protected] https://mail.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - https://mail.zope.org/mailman/listinfo/zope-announce https://mail.zope.org/mailman/listinfo/zope )