[elinks-dev] Re: Re: [ELinks] elinks - pasky: Send Pragma: reload if cache_mode is >= N...

Zas zas at norz.org
Sun Oct 5 05:32:54 PDT 2003


On Sun, Oct 05, 2003 at 01:52:56PM +0200, Petr Baudis wrote:
> Dear diary, on Sun, Oct 05, 2003 at 01:10:00PM CEST, I got a letter,
> where Zas <zas at norz.org> told me, that...
> > On Sat, Oct 04, 2003 at 11:37:49AM -0000, cvs at pasky.ji.cz wrote:
> > > This is an automated notification of a change to the ELinks CVS tree.
> > > 
> > > Author: pasky
> > > Module: elinks
> > >    Tag: <TRUNK>
> > >   Date: Sat Oct  4 11:37:49 2003 GMT
> > > 
> > > ---- Log message:
> > > 
> > > Send Pragma: reload if cache_mode is >= NC_RELOAD. Idea by Radim Kolar <hsn at cybermail.net>.
> > > 
> > > 
> > > ---- Files affected:
> > > 
> > > elinks/src/protocol/http:
> > >    http.c (1.184 -> 1.185) 
> > >     http://cvsweb.elinks.or.cz/cvsweb.cgi/elinks/src/protocol/http/http.c.diff?r1=1.184&r2=1.185&f=u
> > > 
> > > 
> > > ---- Diffs:
> > > 
> > > Index: elinks/src/protocol/http/http.c
> > > diff -u elinks/src/protocol/http/http.c:1.184 elinks/src/protocol/http/http.c:1.185
> > > --- elinks/src/protocol/http/http.c:1.184	Fri Sep 26 23:36:56 2003
> > > +++ elinks/src/protocol/http/http.c	Sat Oct  4 13:37:44 2003
> > > @@ -603,6 +603,11 @@
> > >  		add_to_string(&header, "Cache-Control: no-cache\r\n");
> > >  	}
> > >  
> > > +	if (conn->cache_mode >= NC_RELOAD) {
> > > +		add_to_string(&header, "Pragma: reload\r\n");
> > > +		add_to_string(&header, "Cache-Control: reload\r\n");
> > 
> > Where are these headers defined ??
> > Is there any reference document about them ?
> 
> I didn't find any, but Radim says it is needed for SmartCache

Well, may be SmartCache needs to be fixed...
If we want to provide some workarounds for buggy software then make them
optionnal. 


> and RFC2616 makes further extensions possible. And even if this was ELinks
> own extension

I can't say i agree here. Starting to support our own HTTP extensions is
a non-sense. An http client is made to interact with http servers, and
both need to talk the (almost) same language. Doesn't RFC2616 provides
ways to do the same thing ?

> it IMHO makes sense and could be useful in various conditions.

Which ones ?

> And the future HTTP specifications would be hardly reusing
> the "reload" keyword for anything else than indicating reloading of the
> document, right? Right.

Cache-Control: reload appears in old ietf http spec drafts, but was
forgiven at some time... IMHO, using this in ELinks is not right at all.

--

Zas



More information about the elinks-dev mailing list