#2126: SimFactory should choose an appropriate cfg on a Mac. -------------------------+-------------------------------------------------- Reporter: sbrandt | Owner: Type: defect | Status: reopened Priority: unset | Milestone: ET_2018_08 Component: SimFactory | Version: development version Resolution: | Keywords: -------------------------+-------------------------------------------------- Changes (by rhaas):

* status: review => reopened

Comment:

Please see comments inline in pull request. Cannot be applied as is, and should provide some more robust error checking as well as a statement that this was actually tested by someone (since as far as I know you do not use / or had access to) a mac in the past.

#2126: SimFactory should choose an appropriate cfg on a Mac. -------------------------+-------------------------------------------------- Reporter: sbrandt | Owner: Type: defect | Status: reopened Priority: unset | Milestone: ET_2018_08 Component: SimFactory | Version: development version Resolution: | Keywords: -------------------------+--------------------------------------------------

Comment (by sbrandt):

I just did an experiment. I compiled on a Mac with homebrew using generic.cfg. The obvious problem with doing things this way is that "gcc" is really clang.

However, despite Mac lying about the compiler, Cactus compiled and the test suite ran (except for Memspeed). Given this, I think we can modify the tutorial to remove the Mac-specific instructions and close this ticket.

Ian, could you test to see if this works on a mac also?

#2126: SimFactory should choose an appropriate cfg on a Mac. -------------------------------+--------------------------------- Reporter: Steven R. Brandt | Owner: (none) Type: defect | Status: reopened Priority: unset | Milestone: ET_2018_08 Component: SimFactory | Version: development version Resolution: | Keywords: -------------------------------+---------------------------------

Comment (by Roland Haas):

Looks mostly good. I have added couple of comments to the pull request.

Cannot be pushed to master as is since it contains leftover print statements from debugging it seems. There are some minor things that should be fixed as well and one has to consider what happens on eg debian (and likely Ubuntu) systems where there is a gcc and a gcc-N but gcc is the real thing this time (and points to one of the gcc-N based on either user choice or on the distro settings and gcc is the one that is compatible with most packages provided by the distro while gcc-N may not be).