Laziness leads us to conserve resources, and impatience to "save time" (which is a resource, but of a unique nature).
Pride also leads us to make our work more robust, so that it "always works well."
Thus, two fundamental principles are respected: optimization and robustness.
Is your problem related to various parts of the code being in distinct transactions? Can you try a version having all parts commiting then immediately synchronizing with each other, then SELECT'ing?
I found a fix to the problem, although I never quite pinned down why it was necessary. It wasn't a commit problem, but rather had to do with two objects pointing to the same third object as its "parent", whereas the third object only pointed to one of those other two objects as its "daughter". When queried from either object about the third object's daughter, it gave an answer pointing to that object, despite the parent pointing to just one of the two objects. The fix was to make the non-owned object point to a different owner. But I still am not clear why that was necessary.
Pride also leads us to make our work more robust, so that it "always works well."
Thus, two fundamental principles are respected: optimization and robustness.
Is your problem related to various parts of the code being in distinct transactions? Can you try a version having all parts commiting then immediately synchronizing with each other, then SELECT'ing?