At 12:50 AM +0100 11/28/05, Rafael Garcia-Suarez wrote: >On 11/26/05, Craig A. Berry <[EMAIL PROTECTED]> wrote: >> But there are other structures that are pipe-specific, and it seems >> like a good idea to keep the new sanity check that ensures we close a >> pipe and deallocate its structures from within the same interpreter >> context where it was opened. But I haven't thought of a good way to >> do that for the case where the pipe is opened when there is no >> context at all yet. Any suggestions? > >Looks to me that we should bypass this check when there is no context.
I agree. What is the best way to do that? PL_curinterp == NULL? PL_thr_key == NULL? It looks to me that dTHX translates to pthread_getspecific(), which returns an undefined value before the key is initialized, so we can't depend on dTHX to tell us whether we actually have a thread context or not. The case I am most worried about, though, is the one where we allocate before thread context is initialized and deallocate after (possibly during global destruction). Is global destruction supposed to be smart enough to recognize things that were created before there was a thread context and deallocate them differently? -- ________________________________________ Craig A. Berry mailto:[EMAIL PROTECTED] "... getting out of a sonnet is much more difficult than getting in." Brad Leithauser
