On 31 May 2010 15:01, Paul Ruizendaal <p...@planet.nl> wrote: > > >> I'm really short on time right now, but I will try to help you in > making > >> this a cross-platform patch. I can test on WinXP, Linux and FreeBSD. > Can > >> you test on OS-X? > > > > No, sorry. I can add OpenSolaris to the testing platforms though. > > Well, I'll try to review your patch next weekend and see where the Windows > issues might be.
I see the main one being the use of fork - but in the same way as the standard HTTP server does. I suppose the correct thing to do would be to analyse the win32 http server code. > >> > SCGI and FastCGI also play better with HTTP keep alive and similar > >> > features, as the web server is in a better position from the point of > >> > view of managing the connections. > >> > >> Can you please explain this in detail? I used to be a big fan of > FastCGI, > >> until I figured out that it didn't offer features/advantages that were > >> not available in HTTP/1.1. > > > The main problem with using HTTP proxying is that its difficult for the > web > > server to use HTTP/1.1 to talk to the backend - the server ends up > having > > to parse the HTTP response in order to find the content length (if > provided) > > in order to support keep alive. In the end, most web servers just talk > via > > HTTP 1.0, or don't request keep alive. > > Hmmm... I'm not sure that the above is true for *most* web servers, but > you are certainly correct that ngix uses http/1.0 for backend > communication. Thank you for pointing that out. The biggest issue with HTTP 1.1 for backend communication is long running connections since HTTP doesn't support interleaving. - Owen. (By the way, it seems we managed to accidentally get off-list at some point. I'm gonna bring us back on. Sorry for that.)
_______________________________________________ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
