[WK2][NetworkCache] Add support for "Cache-Control: max-stale" request header

[Chris Dumez]

Add support for "Cache-Control: max-stale" request header:
https://tools.ietf.org/html/rfc7234#section-5.2.1.2

[Radar WebKit Bug Importer]

<rdar://problem/20333296>

[Chris Dumez]

Created attachment 249619 [details]
Patch

[Antti Koivisto]

We are now getting to really obscure corners of the spec. Are there any indications that anyone uses this?

[Antti Koivisto]

Comment on attachment 249619 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=249619&action=review

Well I suppose it is a potentially useful feature even if it is unknown. 

r=me but please consider factoring it differently.

> Source/WebKit2/NetworkProcess/cache/NetworkCache.cpp:168
>      bool hasExpired = age > lifetime;
> +    if (hasExpired && !std::isnan(maxStale)) {
> +        // Request is allowing stale responses.
> +        // https://tools.ietf.org/html/rfc7234#section-5.2.1.2.
> +        if ((age - lifetime) < maxStale)
> +            hasExpired = false;
> +    }

maxStale = std::isnan(maxStale) ? 0 : maxStale;
bool hasExpired = age - lifetime > maxStale;

> Source/WebKit2/NetworkProcess/cache/NetworkCache.cpp:181
> +static bool requestRequiresRevalidation(const WebCore::ResourceRequest& request, double& maxStale)
>  {
>      auto requestDirectives = WebCore::parseCacheControlDirectives(request.httpHeaderFields());
> +    maxStale = requestDirectives.maxStale;

This is a bit strange way to factor this. We should probably combine requestRequiresRevalidation and responseHasExpired since you really need to be looking at both request and response to implement all semantics.

[Chris Dumez]

Created attachment 249732 [details]
Patch

[WebKit Commit Bot]

Comment on attachment 249732 [details]
Patch

Clearing flags on attachment: 249732

Committed r182145: <http://trac.webkit.org/changeset/182145>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.