Ryan and I did discuss this but I think it was face to face at Santa Clara when I made some big changes to that code to get it working :) I'll have a look at the code and see what's going on, but we should be using the offset if there is one, or at least so I believe without reviewing the code.
david > [EMAIL PROTECTED] wrote: > > >I believe this is a bug. However, we have never been able to reliably fix > >it. You should check the APR list archives, because David Reid and I > >discussed this a while back. It was probably around March of this year. > > > Hmm. Couldn't find anything there ... > > >You could also check CVS for the logs for the time code on Unix. That > >would give you a better idea of when we had the discussion. > > > All I saw was something about tm_gmtoff not being present on Solaris. > But there's already a workaround for that, isn't there? > > I fail to see how apr_implode_time() could /ever/ be wrong if it uses > tm_gmtoff. Whether that field is set correctly or not is another matter, > and most probably shows up only on very few platforms. Right now, > apr_implode_time() is /always/ wrong. Surely it sould be a lesser evil > to just fix it?
