Unfortunately, that doesn't solve any of my problems. I would gladly drop hulahop entirely (it's a mess and very low-level), but several people have expressed concern that gecko might be a better choice long-term. However, I have not been able to confirm any of their performance concerns. In my tests, gecko always used more memory and was much slower than webkit. Also, xulrunner is made first for firefox and second for embedding, and it shows painfully. On the other hand, I can't decide by myself to use PyGI since it's too experimental for a platform's main browser and the rest of Sugar doesn't use it. However, the API of pywebkitgtk would be similar to webkitgtk+PyGI, so switching later on shouldn't be very hard. Especially since pywebkitgtk's author himself already did it. Because of various hulahop & xpcom quirks, webwrap (the abstraction layer) is proving increasingly hard to write. I will give it a few more days, but if I can't figure out a clean way to wrab both hulahop and pywebkitgtk I'll drop hulahop entirely and let any future switching back to hulahop rely on git. <blockquote type="cite">On 16 Jun 2010 11:07, "Tomeu Vizoso" <<a href="mailto:tomeu@sugarlabs.org">tomeu@sugarlabs.org</a>> wrote: On Sun, Jun 6, 2010 at 15:33, Lucian Branescu <<a href="mailto:lucian.branescu@gmail.com">lucian.branescu@gmail.com</a>> wrote: > I've received eve...Just wanted to make sure you know that the pywebkitgtk+ maintainer recommends using PyGI instead: <a href="http://janalonzo.wordpress.com/2010/01/18/using-introspected-webkitgtk-in-gwibber/" target="_blank">http://janalonzo.wordpress.com/2010/01/18/using-introspected-webkitgtk-in-gwibber/</a> That was written in January and since then PyGI has progressed greatly. I personally think that you should have less concerns than other people that are moving to PyGI right now. As a general remark, I don't think it's a good idea to cling to software modules whose authors are so willing to drop down and will be less painful in the medium term if people start moving now. That said, it hasn't been decided yet that Sugar will depend on PyGI from 0.90 on (see a new thread from today). Regards, Tomeu > Since it's already late into the project, unless someone has a better > idea, I'll stick to fully...</blockquote>