133445 2014-06-02 07:06:06 -0700 [iOS] WebCore fails to build: platform/network/mac/ResourceHandleMac.mm:729:62: error: use of undeclared identifier '_CFURLConnectionCopyTimingData' 2014-06-02 17:37:55 -0700 1 1 1 Unclassified WebKit Page Loading 528+ (Nightly build) Unspecified Unspecified RESOLVED FIXED InRadar P2 Normal --- 1 ddkilzer ddkilzer achristensen ap dfarler sam oldest_to_newest 1012931 0 ddkilzer 2014-06-02 07:06:06 -0700 Seeing these errors on some internal buildbots: Source/WebCore/platform/network/mac/ResourceHandleMac.mm:729:62: error: use of undeclared identifier '_CFURLConnectionCopyTimingData' Source/WebCore/platform/network/mac/ResourceHandleMac.mm:727:97: error: unused parameter 'timing' [-Werror,-Wunused-parameter] void ResourceHandle::getConnectionTimingData(CFURLConnectionRef connection, ResourceLoadTiming& timing) 2 errors generated. 1012932 1 ddkilzer 2014-06-02 07:06:30 -0700 <rdar://problem/17090035> 1012933 2 232377 ddkilzer 2014-06-02 07:23:53 -0700 Created attachment 232377 Patch v1 1012946 3 ddkilzer 2014-06-02 09:04:10 -0700 This patch built with an iOS Release build internally as well. 1012952 4 232377 ap 2014-06-02 09:17:08 -0700 Comment on attachment 232377 Patch v1 View in context: https://bugs.webkit.org/attachment.cgi?id=232377&action=review I don't understand this patch, what are the bots that fail? Everything was building correctly for a long time. r=meto fix the build anyway. > Source/WebCore/platform/network/mac/ResourceHandleMac.mm:64 > +#else > +typedef struct _CFURLConnection* CFURLConnectionRef; > +extern "C" { > +CFDictionaryRef _CFURLConnectionCopyTimingData(CFURLConnectionRef); > +} > +#endif This shouldn't be an #else. We want to have the declarations even when the private header is included, because this way, we will find out about changes in SPIs sooner and more reliably. It's very difficult to debug issues caused by incorrect SPI declarations. Also, this should be under #if USE(CFNETWORK), to match the code that uses CFNetwork. 1013029 5 ddkilzer 2014-06-02 16:35:09 -0700 (In reply to comment #4) > (From update of attachment 232377 [details]) > View in context: https://bugs.webkit.org/attachment.cgi?id=232377&action=review > > I don't understand this patch, what are the bots that fail? Everything was building correctly for a long time. See internal email thread, or the radar. > r=meto fix the build anyway. > > > Source/WebCore/platform/network/mac/ResourceHandleMac.mm:64 > > +#else > > +typedef struct _CFURLConnection* CFURLConnectionRef; > > +extern "C" { > > +CFDictionaryRef _CFURLConnectionCopyTimingData(CFURLConnectionRef); > > +} > > +#endif > > This shouldn't be an #else. We want to have the declarations even when the private header is included, because this way, we will find out about changes in SPIs sooner and more reliably. > > It's very difficult to debug issues caused by incorrect SPI declarations. > > Also, this should be under #if USE(CFNETWORK), to match the code that uses CFNetwork. I thought we tried to avoid adding headers in #if/#endif blocks unless absolutely necessary. Technically, both this and the __has_include() below it should be in outer #if ENABLE(WEB_TIMING)/#endif macros as well. Do we really want to do that? 1013036 6 ap 2014-06-02 17:07:54 -0700 > I thought we tried to avoid adding headers in #if/#endif blocks unless absolutely necessary. These are not WebKit headers, so I think that it's the right thing to do. With WebKit headers, yes, we try to have the guards inside the headers. 1013040 7 ddkilzer 2014-06-02 17:37:55 -0700 Committed r169551: <http://trac.webkit.org/changeset/169551> 232377 2014-06-02 07:23:53 -0700 2014-06-02 09:17:08 -0700 Patch v1 bug-133445-20140602072328.patch text/plain 1637 ddkilzer U3VidmVyc2lvbiBSZXZpc2lvbjogMTY5NTE5CmRpZmYgLS1naXQgYS9Tb3VyY2UvV2ViQ29yZS9D aGFuZ2VMb2cgYi9Tb3VyY2UvV2ViQ29yZS9DaGFuZ2VMb2cKaW5kZXggYzk3OTFjZTg0N2UyMWEy Nzc1NzEyZjg3Y2YwNTc2MmYwNzkwZGFhOS4uM2I3OTJkOTUwYjc2OGRiN2EyZmU3OTNkYWU4NDhl NmEwY2QxYThmYiAxMDA2NDQKLS0tIGEvU291cmNlL1dlYkNvcmUvQ2hhbmdlTG9nCisrKyBiL1Nv dXJjZS9XZWJDb3JlL0NoYW5nZUxvZwpAQCAtMSwzICsxLDE0IEBACisyMDE0LTA2LTAyICBEYXZp ZCBLaWx6ZXIgIDxkZGtpbHplckBhcHBsZS5jb20+CisKKyAgICAgICAgW2lPU10gV2ViQ29yZSBm YWlscyB0byBidWlsZDogcGxhdGZvcm0vbmV0d29yay9tYWMvUmVzb3VyY2VIYW5kbGVNYWMubW06 NzI5OjYyOiBlcnJvcjogdXNlIG9mIHVuZGVjbGFyZWQgaWRlbnRpZmllciAnX0NGVVJMQ29ubmVj dGlvbkNvcHlUaW1pbmdEYXRhJworICAgICAgICA8aHR0cDovL3dlYmtpdC5vcmcvYi8xMzM0NDU+ CisgICAgICAgIDxyZGFyOi8vcHJvYmxlbS8xNzA5MDAzNT4KKworICAgICAgICBSZXZpZXdlZCBi eSBOT0JPRFkgKE9PUFMhKS4KKworICAgICAgICAqIHBsYXRmb3JtL25ldHdvcmsvbWFjL1Jlc291 cmNlSGFuZGxlTWFjLm1tOiBBdHRlbXB0IHRvIGZpeCB0aGUKKyAgICAgICAgYnVpbGQgYnkgaW5j bHVkaW5nIGEgcHJpdmF0ZSBoZWFkZXIgYW5kIGxvY2FsIGRlY2xhcmF0aW9uLgorCiAyMDE0LTA2 LTAxICBTaW1vbiBGcmFzZXIgIDxzaW1vbi5mcmFzZXJAYXBwbGUuY29tPgogCiAgICAgICAgIERv bid0IHNjcm9sbCB0byBub24taW50ZWdyYWwgc2Nyb2xsIG9mZnNldHMKZGlmZiAtLWdpdCBhL1Nv dXJjZS9XZWJDb3JlL3BsYXRmb3JtL25ldHdvcmsvbWFjL1Jlc291cmNlSGFuZGxlTWFjLm1tIGIv U291cmNlL1dlYkNvcmUvcGxhdGZvcm0vbmV0d29yay9tYWMvUmVzb3VyY2VIYW5kbGVNYWMubW0K aW5kZXggODQ2NjcwNmFmZDYwMWU1NjRkMWFlYmJkNjBlNjJhNjMwMTM3MmU2ZS4uNmVlNzQ4OGJl M2UyZGViNTAzZGVhMzQyZTE3MDEwN2NhM2EzZDdkYSAxMDA2NDQKLS0tIGEvU291cmNlL1dlYkNv cmUvcGxhdGZvcm0vbmV0d29yay9tYWMvUmVzb3VyY2VIYW5kbGVNYWMubW0KKysrIGIvU291cmNl L1dlYkNvcmUvcGxhdGZvcm0vbmV0d29yay9tYWMvUmVzb3VyY2VIYW5kbGVNYWMubW0KQEAgLTU0 LDYgKzU0LDE1IEBACiAjaW1wb3J0IDx3dGYvdGV4dC9CYXNlNjQuaD4KICNpbXBvcnQgPHd0Zi90 ZXh0L0NTdHJpbmcuaD4KIAorI2lmIF9faGFzX2luY2x1ZGUoPENGTmV0d29yay9DRlVSTENvbm5l Y3Rpb25Qcml2Lmg+KQorI2ltcG9ydCA8Q0ZOZXR3b3JrL0NGVVJMQ29ubmVjdGlvblByaXYuaD4K KyNlbHNlCit0eXBlZGVmIHN0cnVjdCBfQ0ZVUkxDb25uZWN0aW9uKiBDRlVSTENvbm5lY3Rpb25S ZWY7CitleHRlcm4gIkMiIHsKK0NGRGljdGlvbmFyeVJlZiBfQ0ZVUkxDb25uZWN0aW9uQ29weVRp bWluZ0RhdGEoQ0ZVUkxDb25uZWN0aW9uUmVmKTsKK30KKyNlbmRpZgorCiAjaWYgX19oYXNfaW5j bHVkZSg8Rm91bmRhdGlvbi9OU1VSTENvbm5lY3Rpb25Qcml2YXRlLmg+KQogI2ltcG9ydCA8Rm91 bmRhdGlvbi9OU1VSTENvbm5lY3Rpb25Qcml2YXRlLmg+CiAjZWxzZQo=