On Mon, 16 Sep 2002, Axel =?iso-8859-1?q?D=F6rfler ?= wrote: > Ingo wrote: > > I think, it is usually desired to compile the tests with `-g'. And `- > > g' > > sort of implies `-O0'. IIRC the BFS tests are the only ones > > recompiling > > production code into the tests, while all other tests just link > > against > > the compiled code being compiled with whatever options desired. > > > > While I'm rethinking the whole debug/warnings/options anyway (I have > > no > > completely satifying solution yet though), I guess, it won't harm to > > introduce a NO_TEST_DEBUG variable to turn off the current behavior > > (i.e. that the test rules override DEBUG and OPTIM). Would that be > > acceptable? > > Yes, I actually think that this is the right way to go :-) Alright, I will add it then. > One shouldn't expect compiler bugs everywhere, and implementation bugs > should be much more common, so we should probably focus on those... Good idea. ;-) > It would be just nice to have an easy way of using the test suites with > a specific optimization level. Well, the question is, how you define `easy way'. As soon as NO_TEST_DEBUG is available, you can use e.g. `NO_TEST_DEBUG=1 OPTIM=-O1 jam ...'. CU, Ingo