On Fri, Apr 6, 2018 at 10:18 AM, Yann Ylavic <ylavic....@gmail.com> wrote: > On Fri, Apr 6, 2018 at 8:57 AM, Ruediger Pluem <rpl...@apache.org> wrote: >> >> I don't think that a remark is strong enough here. I would like to >> see this parameter either added to a new apr_reslist_create_ex or if >> we want to stick with apr_reslist_fifo_set it should return an error >> if elements are already in the reslist. > > r1828492, I kept it as a helper rather than _create_ex() because: > - it's already how settings seem to be added there (timeout, ...) > - _create() has a substantial number of args already :/ > - _create_ex() are definitive extensions or may end up in endless > discussions for the next function name when/should an _ex_ex() is/be > needed :)
Also, the lock is not acquired so the empty test is not that accurate when multi-threaded. I'd prefer to keep it light since we document the expected usage in the first place, and I wouldn't want to penalize respectful users... > > Thanks for the review! > > Regards, > Yann.