Sorry, didn't have time to re-run with updated libraries yet - plan to do it in coming days.
27.12.2016, 01:34, "Faré" fahree@gmail.com:
OK, so apparently:
- make-operation: You need an updated iolib for further tests as this
causes a lot of errors :-(
- make-operation: eco, prove need to use make-operation (sent bug
reports). Many systems use prove. https://github.com/eudoxia0/eco/issues/1 https://github.com/fukamachi/prove/issues/35
- the uffi-tests bug is probably due to the update in cffi
Yes, maybe - it didn't fail before we updated cffi
— dunnowhether it's addressed upstream or not. If not, it needs be reported. Another package complains about an old uffi, so, maybe try to update uffi?
- the failures to locate some .so's might or might not be due to the
CFFI update.
- make-operation: cl-protobufs had the bug, it was fixed already, but
not yet in ql; no system in ql uses it so no need to update for now but if you do we can check that nothing else breaks.
- make-operation: error in lambda-reader-8bit is my fault. Fixed.
- Warning in uiop is my fault. Fixed.
PS: For redirecting the output of subprocesses, you can't just redirect a lisp stream; the Lisp process itself must be started with its outputs properly redirected. You can start the building Lisp process with e.g. (uiop:run-program (lisp-invocation:lisp-invocation ...) :output my-logfile-pathname :error-output :output :input nil)
I doubt I will do it soon: I need process ID of the child process, uiop:run-program doesn't support this. Also I'm not sure passing output file will work reliably on all platforms.