---------- Forwarded message ---------- From: Ming Fu <[email protected]> Date: Thu, Sep 17, 2009 at 10:10 AM Subject: Re: [Polipo-users] A number of suggestions To: Juliusz Chroboczek <[email protected]>
On Wed, Sep 16, 2009 at 5:09 PM, Juliusz Chroboczek < [email protected]> wrote: > > I attached a patch on a few issues. > > Thanks. > > > 1. Do not propagate no-cache from Pragma to Cache-Control. > > Unless I'm mis-understanding something, that's not what your patch > does -- it disables sending of Cache-Control information to the client > altogether. > It stores Cache-Control the same as Cookie, make a string copy and pass them along. Polipo no longer rewrite the cache control line. We keep our interpretation of it for our own use, but pass the original cache control header as it is. See the goto save: line > > > It destroy IE plugins if the pragma have no-cache. > > I'm somewhat undecided on this issue. On the one hand, most modern HTTP > clients are able to parse both Pragma and Cache-Control, so keeping them > separate is not that big a deal. > > On the other hand, IE's bug in this case is a major fuck-up; there is no > excuse for being unable to pass an uncachable object to a plugin. While > I'm quite willing to include hacks in Polipo for working around > reasonable bugs, I'm rather of the opinion that catering to such > outrageous brokenness is a thing no self-respecting computer scientist > ought to do. > > I'll think about it. > Proxy's job is to make the web work. We can't police it. We will have no user if we don't commendation Microsoft. It seems they know the issue for a long time. They see it as a feature rather than bug. So there is no hope Microsoft will change IE behaviour on this issue. > > 2. Do not upgrade 1.0 like Connection: close server reply to 1.1 reply to > > client. > > No, sorry. Right now, Polipo is not a tunnelling proxy; it upgrades > traffic to the capabilities advertised by the client. Behaving as > a tunnelling proxy in some cases but not in others is only adding > confusion to what is already an overly-complex program. > > Adding proper tunnelling capability to Polipo is not an uninteresting > project, but not one I'm at leisure to tackle right now. > > > 3. If Content-Range header is malformed, make it as an unknown header, > not > > an error. Server misses the "bytes" keyword often. > > That's a useful observation, thanks. However, what your patch does > (considering the partial body as the full body) is not the right > solution, as it risks corrupting the cache if the object is cachable. > The right solution is to make the ``bytes'' marker optional. > This is fine too. However, we also regenerate the header and will put the bytes back. Are we sure it has to be bytes. > > > 4. Keep Expire head as is. Keep what our interpret of it only to > > ourselves. > > What is the actual problem you're trying to solve? > The client got throw off guard when the page expired in the 70s when polipo don't understand the expiry day. Pass it the date we don't understand is a better alternative. Maybe these client can parse the date better than we do. This is the same as my Cache-Control change, make a copy and pass along the same as Cookie. Keep our interpretation of the date only to ourselves. Regards, Ming ------------------------------------------------------------------------------ Come build with us! The BlackBerry® Developer Conference in SF, CA is the only developer event you need to attend this year. Jumpstart your developing skills, take BlackBerry mobile applications to market and stay ahead of the curve. Join us from November 9-12, 2009. Register now! http://p.sf.net/sfu/devconf _______________________________________________ Polipo-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/polipo-users
