William A. Rowe, Jr. wrote: > I need to clarify this; > > on unix, with apr_procattr_child_XXX_set() (where XXX is in/out/err), or > with partial apr_procattr_io_set() (where only some of in/out/err are set), > > does apr_proc_create ensure we have the parent processes' std in, out, err > handles where they are not explicitly overridden with the calls I mentioned > above?
With the lack of response, I'll presume that, yes, any weren't filled in are propagated. For Windows, this just means picking up the now-current standard handle (assuming, also, that they are set inherited already). So, next question about apr_procattr_child_[in|out|err]_set(attr, file, NULL). I'll presume from the sources that this is completely legitimate? No need to actually have a parent handle/fd, since we check if it's null and take no action relative to the parent, if it wasn't defined. Bill
