* Dave Mitchell <davem@iabyn.com> [2016-05-03T11:56:31] > [...] Coro breakage [...] Below is some recent chatter from #p5p. I have edited it only for comprehensibility and brevity. <Nicholas> leont: here's the bit I don't get. If I comment out the assignments in State.xs that monkey patch in the overrides (so that no overriding is happening) all the tests still pass <leont> … <leont> Nicholas: we have all been idiots <Nicholas> because? <leont> For not seeing this earlier :-/ <Nicholas> 1) there are no tests for the actual deep bug that someone hit (see comment in State.xs "The original one doesn't provide for reading back of PL_diehook/PL_warnhook") <Nicholas> 2) maybe it's all for working around bugs that are already fixed by 5.22.0 time <Nicholas> 3) the only thing that it's trying to cope with is actually perlio.c does of directly localising setting PL_warnhook. In which case, fix perlio.c to use (I think) save_helem_flags() <leont> 5.22.2 + PadlistNAMES fix works fine for me too <leont> the tests [...] do actually test what they need to test <leont> In particular t/13_diewarn.t <Nicholas> there are two or three possible reasons So... what's the deal? -- rjbsThread Previous | Thread Next