On 07/11/2016 10:18 AM, Christos Tsantilas wrote: > This patch includes a Downloader class which implemented as independent > AsyncJob class (in the initial patch was a ConnStateData kid). > > Currently runs an other related discussion under the mail thread "Care > and feeding of ConnStateData", but I believe that this discussion does > not affect a decision about applying or not the Downloader class.
I agree. The "care and feeding" thread uses the latest stand-alone Downloader design to illustrate class boundaries/relationships, and the latest Downloader uses some ideas from that thread as well, but the _disagreements_ on that thread do not affect the latest Downloader class design. This long-awaited "fetch missing certificates" project is not blocked on that thread. > + // Do not check for redirect, tos,nfmark and sslBump > + http->calloutContext->redirect_done = true; > + http->calloutContext->tosToClientDone = true; > + http->calloutContext->nfmarkToClientDone = true; > + http->calloutContext->sslBumpCheckDone = true; Why do we not want to do each of those things? I suspect the answer will differ for each feature, so we probably should document those reasons separately. For example: // no support for URL rewriting (yet) because ... http->calloutContext->redirect_done = true; // no Squid-client markings because there is no client connection // TODO: Make doCallouts() smarter about client connection presence? http->calloutContext->tosToClientDone = true; http->calloutContext->nfmarkToClientDone = true; // no need to bump because Squid is the true client here http->calloutContext->sslBumpCheckDone = true; This feature is useful even without URL rewriting support so we do not have to support URL rewrites, especially if adding such is not trivial, but I think we need to disclose why we are disabling things. Alex. _______________________________________________ squid-dev mailing list [email protected] http://lists.squid-cache.org/listinfo/squid-dev
