Raymond> Grabbed a tarball and built it from scratch. Everything Raymond> seems fine after an hour of general usage. One issue: I Raymond> was using --enable-error-checking=none previously. This Raymond> seems to recognize that as being the same as Raymond> --with-error-checking, but "none" is not recognized. Stephen> Thanks for the report. This is known broken due to a change in Stephen> implementation and semantics of default handling. This was Stephen> inadvertant. I thought Malcolm has already checked in a fix, but Stephen> apparently not. I expect it momentarily, Christmas and all that Stephen> interrupts things. Raymond> Does that mean I have to explicitly list all of the valid Raymond> tests, prefixed by "none" or explicitly say noextents, Raymond> nobufpos, etc.? Stephen> IIRC "none" is completely ignored; you need to explicitly list all the Stephen> ones you want to shut off, each prefixed by "no". Yes, I'll have to do that, which is ok. I should have read the mailing lists before sending this message out. :-) Now I get to reconfig and rebuild to turn off error checking; GNUS is just too annoyingly slow with all the error checking on. Thanks, Ray