Hi Ian, Ian McWilliam wrote on Wed, Sep 06, 2017 at 11:31:21PM +0000:
> You are correct. spell hadn't at that point been updated > and still pointed to the old libc. Thanks for confirming, so nothing is broken in libc. Phew. > What's interesting is a > make repackage update in textproc/aspell directory sees the failure. > make repackage update in textporoc/aspell/core first, solved the issue. I'm sorry, i can't help with that. Even though it is partially documented, i never managed to understand the maze of targets in bsd.port.mk(5). It is overengineered and of excessive complexity to the point that documenting it no longer helps using it. Yes, this is one of the few areas where the historically grown, rampant design of a subsystem ran riot and is now strongly at variance with the OpenBSD project goals of simplicity, usability, and easy debugability. OpenBSD user interfaces ought to be small. My personal way of dealing with the mess is to practically never use any targets except - makesum - show - extract - patch - update-patches - build - fake - update-plist - package - clean I always figure out the dependencies by hand and manually build all ports in the correct order. I exclusively use pkg_add(1) for installing ports, even those that i build myself, and almost never let "make" install any ports for me. Personally, i wouldn't object to deleting most other targets (including repackage, update, and most others), to drain the swamp and facilitate debugging. But i expect that's a minority opinion. In particular, bulk builders and prolific porters are likely to disagree. Yours, Ingo
