After lots of attempts at changing the the to be pickled contents, I
modified the columntype of value to be longtext instead of blob.
This solved all caching problems.
That's rather odd, perhaps the MySQL connector is doing something
funky with the formatting when its a blob that isn't done when its
longtext. Did changing the column type also solve the cut off issue?
Yes. I suppose that the problem was the cut off issue. I am not
actually caching the html body. I am storing the entire response
object like beaker_cache does. A Response object whose pickled
representation is truncated can not be restored to its original form
and is deemed to be insecure by the unpickling code.
I don't know whether a lot of people are using beaker.ext.database.
There are not a lot of docs and a lockfile is still required (not sure
why), but it works for me as an easy to implement and more importantly
completely tranparent caching mechanism.
Bob
6 messages in com.googlegroups.pylons-discussRe: Beaker + SQLAlchemy backend problem