On 2009/11/22 16:49:17, jat wrote: > However, I think perhaps disconnect callbacks should be moved to the common code > and added to SessionHandler. That would still leave noticing disconnects on > sends, but it might be worth the effort. What do you think?
Sure, I can give this a shot. Common code sounds better, the only thing is I'll need to add a virtual method to SessionHandler so the implementation can be common. I guess I can it make it non-pure virtual for now (so we don't break other platforms) with a TODO to make it pure virtual when all of the plugins have implemented it. > I haven't tested this -- I assume you will with all supported browsers before > committing. Don't forget to update the binaries as well. So far I've only changed IE and Chrome. I've tested IE6 and Chrome, but need someone to test IE7 & IE8. http://gwt-code-reviews.appspot.com/112801/diff/1007/13 File plugins/ie/oophm/oophm/JavaObject.cpp (right): http://gwt-code-reviews.appspot.com/112801/diff/1007/13#newcode155 Line 155: bool isException(false); Done. http://gwt-code-reviews.appspot.com/112801/diff/1007/18 File plugins/npapi/prebuilt/gwt-dev-plugin/manifest.json (right): http://gwt-code-reviews.appspot.com/112801/diff/1007/18#newcode3 Line 3: "version": "1.0", I wasn't planning to commit this change yet, but I'm not sure why it was failing to parse what's there. Alternatively, we could go with a very different version scheme.... like "2009.11.22" and not care about artificial version stamps. Should I try this? http://gwt-code-reviews.appspot.com/112801 -- http://groups.google.com/group/Google-Web-Toolkit-Contributors
