Mark, Hmm, just realized, to add on top of what you said, method has a major thread-safety problem if the bytevector output port is accessed from more than one thread. Definitely need to fix this. Freja Nordsiek On Mon, Jun 19, 2017 at 8:03 AM, Mark H Weaver wrote: > Freja Nordsiek writes: > > > Also, I found one error in the get-output-bytevector procedure in > > (scheme base), which was that the procedure discarded the bytes > > already written. get-output-bytevector is not supposed to be > > destructive to the bytes already written, but the R6RS output > > bytevector reading procedures are destructive. I made a patch with a > > very simple fix, which is to just write the bytes back. > > Ah, good catch! I'd like to fix this in a different way though: instead > of removing the bytes and then adding them back, I'd prefer to modify > r6rs-ports.c to support this (read-only) operation directly. > > Thanks, > Mark >