build-gtkdoc is currently for WebKit1 only and does not integrate into build-webkit and "make docs".
Created attachment 115609 [details] Patch
Created attachment 115729 [details] Patch
Thanks for the patch. If this patch contains new public API please make sure it follows the guidelines for new WebKit2 GTK+ API. See http://trac.webkit.org/wiki/WebKitGTK/AddingNewWebKit2API
Comment on attachment 115729 [details] Patch Attachment 115729 [details] did not pass gtk-ews (gtk): Output: http://queues.webkit.org/results/10508494
This patch fails to build because the bots need a clean build.
Comment on attachment 115729 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=115729&action=review Does this unconditionally run gtk-doc without generating the html for every build? > Source/WebCore/GNUmakefile.list.am:-2379 > - Source/WebCore/page/MouseLockable.h \ This looks unrelated, is it a distcheck build fix? > Source/WebCore/GNUmakefile.list.am:2404 > + Source/WebCore/page/PointerLock.h \ Ditto. > Source/WebKit/gtk/docs/GNUmakefile.am:-105 > -include $(top_srcdir)/Source/WebKit/gtk/GNUmakefile.gtk-doc.am I guess you should remove GNUmakefile.gtk-doc.am too. > Tools/GNUmakefile.am:235 > +docs: $(BUILT_SOURCES) \ > + libwebkitgtk-@WEBKITGTK_API_MAJOR_VERSION@.@WEBKITGTK_API_MINOR_VERSION@.la \ > + libwebkit2gtk-@WEBKITGTK_API_MAJOR_VERSION@.@WEBKITGTK_API_MINOR_VERSION@.la > + $(srcdir)/Tools/gtk/generate-gtkdoc > + Does this only happen when using make docs? We should generate the docs when running make distcheck to include the html docs in the tarball, so I guess we should add doc-dist-hook. And also rules to clean it up. We still have GTK_DOC_CHECK macro in configure, which adds --enable-gtk-doc, --enable-gtk-doc-html and --enable-gtk-doc-pdf, I think we should respect them, at least --enable-gtk-doc, because everybody expects it to work in a project using gtk-doc.
(In reply to comment #6) > (From update of attachment 115729 [details]) > View in context: https://bugs.webkit.org/attachment.cgi?id=115729&action=review > > Does this unconditionally run gtk-doc without generating the html for every build? This happens for builds run via build-webkit. This is so that the EWS will fail if a patch introduces documentation warnings. > > > Source/WebCore/GNUmakefile.list.am:-2379 > > - Source/WebCore/page/MouseLockable.h \ > > This looks unrelated, is it a distcheck build fix? > > > Source/WebCore/GNUmakefile.list.am:2404 > > + Source/WebCore/page/PointerLock.h \ > > Ditto. Oops. webkit-patch snuck those in. These were make distcheck fixes I needed to test 'make dist'. > > > Source/WebKit/gtk/docs/GNUmakefile.am:-105 > > -include $(top_srcdir)/Source/WebKit/gtk/GNUmakefile.gtk-doc.am > > I guess you should remove GNUmakefile.gtk-doc.am too. Yep! > > > Tools/GNUmakefile.am:235 > > +docs: $(BUILT_SOURCES) \ > > + libwebkitgtk-@WEBKITGTK_API_MAJOR_VERSION@.@WEBKITGTK_API_MINOR_VERSION@.la \ > > + libwebkit2gtk-@WEBKITGTK_API_MAJOR_VERSION@.@WEBKITGTK_API_MINOR_VERSION@.la > > + $(srcdir)/Tools/gtk/generate-gtkdoc > > + > > Does this only happen when using make docs? We should generate the docs when running make distcheck to include the html docs in the tarball, so I guess we should add doc-dist-hook. And also rules to clean it up. We still have GTK_DOC_CHECK macro in configure, which adds --enable-gtk-doc, --enable-gtk-doc-html and --enable-gtk-doc-pdf, I think we should respect them, at least --enable-gtk-doc, because everybody expects it to work in a project using gtk-doc. After talking with Carlos via IRC, it seems that adding --enable-gtk-doc is quite useful.
Created attachment 115835 [details] Properly handle --enable-gtk-doc
Comment on attachment 115835 [details] Properly handle --enable-gtk-doc Attachment 115835 [details] did not pass gtk-ews (gtk): Output: http://queues.webkit.org/results/10519167
Created attachment 116444 [details] Patch
Created attachment 116493 [details] Updated patch based on Carlos' comments
Created attachment 116498 [details] Fix a couple bugs
Comment on attachment 116498 [details] Fix a couple bugs View in context: https://bugs.webkit.org/attachment.cgi?id=116498&action=review Nice patch! r- mostly because of the webkit-build-directory issue > Tools/GNUmakefile.am:260 > + @rm -rf Documentation/webkitgtk Documentation/webkit2gtk > + -@rmdir Documentation Can that be replaced with rm -rf Documentation? > Tools/gtk/generate-gtkdoc:58 > + process = subprocess.Popen(['perl', script_path('webkit-build-directory'), '--configuration', 'gtk'], > + stdout=subprocess.PIPE) This will always return the path to a Release build because we never call set-webkit-configuration :(
Created attachment 116516 [details] No longer use webkit-build-directory, work with _build, move utilities to common.py
Created attachment 116519 [details] Patch
Created attachment 116520 [details] Correct patch
Comment on attachment 116519 [details] Patch Thanks!
Comment on attachment 116519 [details] Patch Oops
Comment on attachment 116520 [details] Correct patch Thanks :)
Comment on attachment 116520 [details] Correct patch View in context: https://bugs.webkit.org/attachment.cgi?id=116520&action=review > Tools/gtk/common.py:57 > + build_dir = top_level_path() This should be build_dir = top_level_path('WebKitBuild') for the cases where webkit-build is not used and there isn't Release nor Debug dirs > Tools/gtk/common.py:62 > + sys.exit(1) There's no import sys in this file, this will fail.
Committed r101174: <http://trac.webkit.org/changeset/101174>
It appears that this patch broke GTK build: http://build.webkit.org/builders/GTK%20Linux%2064-bit%20Debug/builds/27884 http://build.webkit.org/builders/GTK%20Linux%2064-bit%20Debug/builds/27884/steps/compile-webkit/logs/stdio make: *** No rule to make target `../../Source/WebKit/gtk/docs/GNUmakefile.am', needed by `../../GNUmakefile.in'. Stop. Failed to build WebKit using 'make'! program finished with exit code 2 elapsedTime=6.273611