Mike Kupfer wrote: >>>>>> "jmcp" == James C McPherson <[EMAIL PROTECTED]> writes: > > jmcp> - resolve 6414851, so that if -D is set in your env file, then > jmcp> bldenv does not ignore it, and > > I originally had some concerns about the case where NIGHTLY_OPTIONS has > -D but not -F (i.e., nightly(1) will do both debug and non-debug > builds). It wasn't obvious to me that debug is the right build type for > bldenv to set up in that situation. > > But thinking about it this morning, it occurs to me that the expected > scenario is that the user runs nightly(1), finds a problem (or makes > some changes), and wants to rebuild some subset of the tree. If > nightly(1) is doing both debug and non-debug builds, it does the > non-debug build first and then the debug build, leaving the workspace > with debug binaries. So having bldenv set up a debug build in that case > does seem to be the right thing.
Which is exactly how I try to work and why I stopped using ws in preference to 'bldenv -d env' > jmcp> - ensure that -i on the command line does not turn off lint if > jmcp> -l is specified in your env file. > > Yeah, the discrepancy between -i on the command line and -i in > NIGHTLY_OPTIONS is awkward. Probably better to add support for "+l" to > disable linting. I thought that but didn't want to suggest + options because we didn't have them before (plus they look strange to me :-)). -- Darren J Moffat _______________________________________________ tools-discuss mailing list [email protected]
