On 7/19/12 Jul 19 -9:46 AM, Brandon Allbery wrote: > On Wed, Jul 18, 2012 at 11:58 PM, Robert P. Goldman <[email protected] > <mailto:[email protected]>> wrote: > > That fasl (fast load) file is not the problem. It is recompiled as > part of the testing process, so if it still references the stale > libffi, then the ECL compiler is still referencing it, despite the > port upgrade. So the port upgrade is not sufficient to fix the > problem. I don't know enough about the mechanics to understand why > the command Brandon supplied did not trigger a rebuilding of the ECL > compiler, but it didn't. > > > Or it did but that wasn't enough. It occurs to me that, if it's > generating FASLs as shared objects, it may have a hardcoded reference > somewhere in the source code (a string, not a linker reference that > would be updated automatically) and a patch would be needed to update that.
My mistake. Here's the problem: previous failures mean that the test code did not run to completion. Because it didn't run to completion, the test code *DID NOT WIPE OUT ITS CACHE OF FASLs*. Hence it was loading stale fasls. This is a bug in the test-running code, which I will report to ASDF. At any rate, when I did rm -rf asdf/tmp/, the tests ran. Thanks for your help, R _______________________________________________ macports-users mailing list [email protected] http://lists.macosforge.org/mailman/listinfo/macports-users
