Specific issue was seen from: void Connection::handleWriteResponse(ReplyPtr reply, const asio_error_code& e, std::size_t bytes_transferred) called from a boost call back function. Code review shows that must have been from TcpCon...Koen Deforche
Hey Bruce, Sounds plausible that indeed the long connection times cause issues with prepared statement planning being held. I think reducing the connection lifetime is not such a bad solution --- the pool allows to avoid opening/closing...Koen Deforche
Roel, Seems reasonable to me if the same thing didn't work in Wt 3 neither? Otherwise I would be curious how Wt3 got to react to resize? Regards, koen Koen Deforche
Michael, We do not actually rollback changes you made to in-memory objects ourselves since the intent of the transaction is to bring the database persisted state in sync with the in-memory state. If that fails, that should not necessari...Koen Deforche
Hey, The solution I would propose is to deploy the app at "/" of your server, and indicate in the docroot all static resources resources (---docroot='/docroot;/css,/resources ...'). Then everything that does not match a static resource...Koen Deforche
Most databases do not support unsigned integer types (and the standard doesn't specify them), asfaik. The design of the Wt::Dbo is to model the data that can be stored in the database using appropriate types in C so that a one-on-one rel...Koen Deforche
Hey, Wt::Dbo only takes a call to 'modify()' as an indication that you are modifying the object. But it will flush dirty objects whenever a query needs to be done (to make sure dirty changes are consistent with the result of the query)....Koen Deforche
Hey, The design is so that in a function, of which you don't want to assume how and when it will be called, and if that function requires a transaction, it should use one. At least that makes the function work in all situations. If the ...Koen Deforche