| Code in a TransactionListener which tried to make a backend call would crash because the Invocation context was not setup. Some simple backend calls might be safe, but we really don't want any user code calling into the backend when we're at transaction end, so prevent all access. |
|
More...
|
about 17 years ago
|
| When executing a statement, we want to make sure any previous state was cleaned up, so the code was calling close(). This additionally set a closed flag which would throw exceptions later saying that you can't call various methods because the Statement is closed. It's not actually closed, so split the internal cleanup routine to clear() and make close() something that only a user should call. |
|
More...
|
over 17 years ago
|
| For a multi-threaded pljava function we need to adjust stack_base_ptr before calling into the backend to avoid stack depth limit exceeded errors. Previously this was done only on query execution, but we need to do it on iteration of the ResultSet as well. |
|
More...
|
over 17 years ago
|
| Release 1.4.0. |
|
More...
|
almost 18 years ago
|
| Starting with 8.2, the server reference counts some TupleDescs, so we need to let it know when we are done with it. |
|
More...
|
almost 18 years ago
|
| For a server built with integer datetimes, conversion from SQL to Java timestamps produced bogus results as the result of overflow. Prior to conversion to milliseconds, cast the difference in PG and Unix epochs to a 64 bit type. |
|
More...
|
almost 18 years ago
|
| Recent security releases have altered the API for SetUserId to include an additional parameter indicating if we're in a security definer context. |
|
More...
|
almost 18 years ago
|
| To build against release candidates, the version number parser must know about the RC symbol. |
|
More...
|
almost 18 years ago
|
| When doing batch execution with batches that have different types, the server crashes because it did not replan the query for the new types. Save the type information for each batch and compare it with the previous values to see if we need to replan. |
|
More...
|
about 18 years ago
|
| Batch execution of PreparedStatements failed because addBatch was only saving off the parameter values while it ignored the parameter types. Save both the types and values. |
|
More...
|
about 18 years ago
|
| When deploying a jar file with a deployment descriptor, the deployment failed if the default schema already had a classpath set. A typo was setting the current schema value to the classpath, instead of the default schema. When trying to set the new classpath this bailed out with "No such schema: <classpath>". |
|
More...
|
about 18 years ago
|
| Allow pljava built with JDK < 1.6 to successfully run its tests under a 1.6 JVM by handling unimplemented JDBC4 methods that reflection finds. |
|
More...
|
about 18 years ago
|
| In 8.3 DLLIMPORT has been renamed to PGDLLIMPORT and global_timezone has been renamed to session_timezone. |
|
More...
|
over 18 years ago
|
| 8.3's find_coercion_pathway API has reverted to the old pre-8.3 API, so the changes made earlier to support 8.3 builds are no longer correct or needed. |
|
More...
|
over 18 years ago
|
| Map i486 and i586 $(cpu_host)s to JRE_CPUs of i386. |
|
More...
|
over 18 years ago
|
| Allow building with compilers that don't support unnamed struct and union members by providing names for them. |
|
More...
|
over 18 years ago
|
| Ensure arrayCoerce is initialized when used on pre 8.3 servers. |
|
More...
|
over 18 years ago
|
| Had wrong version check here. We only want SET_VARSIZE for 8.3+. |
|
More...
|
over 18 years ago
|
| To build pljava we need pg_config on the path, not pg_ctl. |
|
More...
|
over 18 years ago
|
| Make pljava build against a 8.3dev server. |
|
More...
|
over 18 years ago
|
| Added missing include of xact.h |
|
More...
|
over 19 years ago
|
| Fixed some issues with Meta-data |
|
More...
|
over 19 years ago
|
| Fixed stack backtrace on failure to load class |
|
More...
|
over 19 years ago
|
| Preparation for 1.3.0 final |
|
More...
|
over 19 years ago
|
| Added PG_MODULE_MAGIC declaration required from PosgreSQL 8.2 and up. |
|
More...
|
over 19 years ago
|
| Fixed bug 1558 |
|
More...
|
over 19 years ago
|
| Prep for 1.3.b2 |
|
More...
|
over 19 years ago
|
| Removed some excess logging |
|
More...
|
over 19 years ago
|
| Fixed some backward compatibility issues Added explicit coercion capability |
|
More...
|
over 19 years ago
|
| PG_CONFIG now overridable Fixed bug concerning self referencing when building with GCJ |
|
More...
|
over 19 years ago
|
