Check for supported schemes and methods.

Check for supported schemes and examine request methods at key points. We support http, https, and file (dbg only) URLs for now.

* Added IsSchemeSupported, IsMethodSupported and IsMethodAndSchemeSupported helpers, and string constants.

* Check for supported schemes and methods during cache selection and during request interception. Must be GET for cache selectino, GET or HEAD for request interception.

* Renamed some data members in WebApplicationCacheHostImpl to more closely match naming elsewhere.

* Added AppCacheHost::Observer to make life easier. (I like the observer model, and even noticed that the chrome code base has a multi-threaded version too (ala Gears)... nice :)

* Switched to using the observer model in AppCacheRequestDispatcher instead of a WeakPtr. One of the observable methods is OnDestructionImminent(host).

* Added gyp dependency on the net library

BUG=none
TEST=none


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=26537

[michaeln, 2009-09-16 07:28:09]

http://codereview.chromium.org/205017/diff/1/9
File webkit/appcache/appcache_host.cc (right):

http://codereview.chromium.org/205017/diff/1/9#newcode178
Line 178: if (associated_cache() || is_selection_pending())
note to self:
if ((cache() && is_complete()) || is_selection_pending())

[michaeln, 2009-09-16 17:25:57]

I have to look into that try failure. While i'm at it, i'll look into holding
the request_context_ ref in AppCacheService too. I want to flip our enable flag
for test_shell.

[michaeln, 2009-09-16 17:41:13]

http://codereview.chromium.org/205017/diff/1/2
File webkit/appcache/web_application_cache_host_impl.cc (right):

http://codereview.chromium.org/205017/diff/1/2#newcode69
Line 69: std::string method = request.httpMethod().utf8();
DCHECK(IsAllUpperCase(method));
is_get_method_ = (method == kHttpGetMethod);

http://codereview.chromium.org/205017/diff/1/2#newcode106
Line 106: if (is_scheme_supported_ && is_get_method_) {
note to self: also move same origin detection to this side of the wire so we
don't buffer/passalong master_entry response data unnecessarily

http://codereview.chromium.org/205017/diff/1/4
File webkit/appcache/web_application_cache_host_impl.h (right):

http://codereview.chromium.org/205017/diff/1/4#newcode6
Line 6: #define WEBKIT_APPCACHE_WEB_APPLICATION_CACHE_HOST_IMPL_H_
Doh... i look at this test case...

RenderViewTest_OnLoadAlternateHTMLText_Test

	scoped_ptr<AppCacheDispatcher>::get [0x010D9C01+17]
(c:\b\slave\win\build\src\base\scoped_ptr.h:96)
	RenderThread::appcache_dispatcher [0x010D848C+28]
(c:\b\slave\win\build\src\chrome\renderer\render_thread.h:110)
	RendererWebKitClientImpl::createApplicationCacheHost [0x010D8410+64]
(c:\b\slave\win\build\src\chrome\renderer\renderer_webkitclient_impl.cc:157)
	url_canon::RawCanonOutputT<char,2048>::`scalar deleting destructor'
[0x02430DBA+3498]
	url_canon::RawCanonOutputT<char,2048>::`scalar deleting destructor'
[0x02430C8F+3199]
	WebCore::MainResourceLoader::load [0x021927CC+156]
(c:\b\slave\win\build\src\third_party\webkit\webcore\loader\mainresourceloader.cpp:488)
	WebCore::premultipliedARGBFromColor [0x01CACB4A+87898]
	WebCore::ImageSource::frameIsCompleteAtIndex [0x01B4D2F6+178390]
	WebCore::ImageSource::frameIsCompleteAtIndex [0x01B4F9E9+188361]
	WebCore::ImageSource::frameIsCompleteAtIndex [0x01B4F726+187654]
	WebCore::ImageSource::frameIsCompleteAtIndex [0x01B4F1DD+186301]
	WebCore::ImageSource::frameIsCompleteAtIndex [0x01B4983E+163358]
	WebCore::ImageSource::frameIsCompleteAtIndex [0x01B495F0+162768]
	WebCore::ImageSource::frameIsCompleteAtIndex [0x01B490EC+161484]
	WebFrameImpl::loadData [0x01023FF9+393]
(c:\b\slave\win\build\src\webkit\glue\webframe_impl.cc:797)
	WebFrameImpl::loadHTMLString [0x010244F5+101]
(c:\b\slave\win\build\src\webkit\glue\webframe_impl.cc:814)
	RenderView::OnLoadAlternateHTMLText [0x010AC4C0+336]
(c:\b\slave\win\build\src\chrome\renderer\render_view.cc:749)
	RenderViewTest_OnLoadAlternateHTMLText_Test::TestBody [0x0044F1D3+147]
(c:\b\slave\win\build\src\chrome\renderer\render_view_unittest.cc:42)

[jennb, 2009-09-16 22:04:05]

http://codereview.chromium.org/205017/diff/1/5
File webkit/appcache/appcache_interfaces.cc (right):

http://codereview.chromium.org/205017/diff/1/5#newcode22
Line 22: url.SchemeIsFile();
should this be 4 space indent? not sure if the indent should be treated like an
if-clause condition or not.  Ditto for line 31.

http://codereview.chromium.org/205017/diff/1/6
File webkit/appcache/appcache_interfaces.h (right):

http://codereview.chromium.org/205017/diff/1/6#newcode75
Line 75: // Useful string constants.
Might want to add a note explaining these are defined in url_constants, but is
redefined here because we don't want to depend on chrome lib.

http://codereview.chromium.org/205017/diff/1/7
File webkit/appcache/appcache_request_handler.cc (right):

http://codereview.chromium.org/205017/diff/1/7#newcode39
Line 39: // If the resource is not to be fetched using the HTTP GET mechanism or
Is FILE an equivalent of HTTP GET?  File means something is already available
offline, so it shouldn't be in the manifest...

[jennb, 2009-09-17 00:38:37]

LGTM

Pending trybot failure resolution and other minor issues, FILE behind debug-only
code. Don't enable flag in this CL.