http://gwt-code-reviews.appspot.com/1469803/diff/6009/plugins/npapi/LocalObjectTable.h File plugins/npapi/LocalObjectTable.h (right):
http://gwt-code-reviews.appspot.com/1469803/diff/6009/plugins/npapi/LocalObjectTable.h#newcode166 plugins/npapi/LocalObjectTable.h:166: set( id, jsObject ); On 2011/07/07 15:19:54, conroy wrote:
nit: extra whitespace here
done... trying to break old habits which include memorized key-presses. http://gwt-code-reviews.appspot.com/1469803/diff/6009/plugins/npapi/NPVariantUtil.h File plugins/npapi/NPVariantUtil.h (right): http://gwt-code-reviews.appspot.com/1469803/diff/6009/plugins/npapi/NPVariantUtil.h#newcode8 plugins/npapi/NPVariantUtil.h:8: // Sun's cstring doesn't define strlen/etc On 2011/07/07 15:19:54, conroy wrote:
given that we don't actually build solaris, we should remove any such
ifdefs
from our code as they are sure to grow stale.
please remove commented out includes as well.
Done. http://gwt-code-reviews.appspot.com/1469803/diff/6009/plugins/npapi/ScriptableInstance.cpp File plugins/npapi/ScriptableInstance.cpp (right): http://gwt-code-reviews.appspot.com/1469803/diff/6009/plugins/npapi/ScriptableInstance.cpp#newcode356 plugins/npapi/ScriptableInstance.cpp:356: void ScriptableInstance::testJsIdentity(const NPVariant* args, unsigned argCount, NPVariant* result) { On 2011/07/07 15:19:54, conroy wrote:
i was thinking more along the lines of checking the UA during the
connect.
there's no need to add a separate API hook to the JS for this.
don't we risk a false positive with this approach?
I talked to John last night before leaving and he suggested against connect. This API is called only when a tab triggers on a gwt-codeserver URL. I've tested all of the changes with 12,13,14(dev) and 14(svn). This test is part of the GWT JUnits that failed on all but the 14(svn). http://gwt-code-reviews.appspot.com/1469803/diff/6009/plugins/npapi/ScriptableInstance.h File plugins/npapi/ScriptableInstance.h (right): http://gwt-code-reviews.appspot.com/1469803/diff/6009/plugins/npapi/ScriptableInstance.h#newcode91 plugins/npapi/ScriptableInstance.h:91: static bool jsIdentitySafe; On 2011/07/07 15:19:54, conroy wrote:
does this really need to be static? it should be fine to test/set this
per
connect
The reason I think connect() was de-emphasized was the changing of the generated hosted.html for each project, and thus every plugin's connect(). http://gwt-code-reviews.appspot.com/1469803/diff/6009/plugins/npapi/prebuilt/gwt-dev-plugin/background.html File plugins/npapi/prebuilt/gwt-dev-plugin/background.html (right): http://gwt-code-reviews.appspot.com/1469803/diff/6009/plugins/npapi/prebuilt/gwt-dev-plugin/background.html#newcode69 plugins/npapi/prebuilt/gwt-dev-plugin/background.html:69: plugin.testJsIdentity( idObject, idObject ); On 2011/07/07 15:19:54, conroy wrote:
if you go the route I suggested of simply testing (or sniffing the UA)
during
connect, then you can remove this chunk.
I was using the NPAPI UA before, but it got a little messy with guessing which build was the inception. I also don't know if they are planing on back porting the change to 13. It was suggested that the test would be simplest, and remove version tracking of the browsers that pass. http://gwt-code-reviews.appspot.com/1469803/ -- http://groups.google.com/group/Google-Web-Toolkit-Contributors