You can control the precedence of jar vs. project.
The build path editor has an "order and export" tab. There you can move things up/down the list. By default, the jars are all at the top, and dependent projects at the bottom. But you can re-order this. I have had to do this when juggling using the cross-test rig, vs. using regular daffodil for testing schemas ________________________________ From: Sloane, Brandon <[email protected]> Sent: Wednesday, May 22, 2019 11:31:23 AM To: [email protected] Subject: Re: Differing behaviour between CLI and SAPI That seems to be it. I verified that my CLI was using the local build, but forgot to verify that my project was using the local one as well. It appears that if you add an eclipse project as a dependency, and have a jar on the build path that provides the same classes, the jar takes precedence. ________________________________ From: Beckerle, Mike <[email protected]> Sent: Wednesday, May 22, 2019 11:20:18 AM To: [email protected] Subject: Re: Differing behaviour between CLI and SAPI Wild guess: you are not linking against the same daffodil jars that the CLI is using? ________________________________ From: Sloane, Brandon <[email protected]> Sent: Wednesday, May 22, 2019 11:08:41 AM To: [email protected] Subject: Differing behaviour between CLI and SAPI I am having an issue where running with the CLI is producing correct results, whereas the SAPI is producing incorrect results. I suspect that this is related to the bitorder unparse error that was recently fixed, as it occurs on unparse and goes away when I switch to "read-only" schema that has had its outputValueCalcs removed. Any insight into where the source of this discrepency might be? Brandon T. Sloane Associate, Services [email protected] | tresys.com
