gtk+-3.0 and something you _didn't_ already know

[technosaurus]

As many of you may have heard, gtk+-3.0 has been released and is _supposed_ to abandon the x11 bindings for cairo (I'll see if it builds with cairo-xcb, but I'm not expecting that to work based on my last source audit) There are several announcements pertaining to this around your favorite nerd news sites, so I won't reiterate the obvious, but I did want to bring everyone's attention to something that could be even better than abandoning X11.

The gtk+-3.X branch now has a separate branch with an experimental html5 backend called "broadway". There is an article here:
http://www.phoronix.com/scan.php?page=n ... &px=ODgzMA
Its a pretty ingenious idea really that could bring cloud computing to the next level. If it is written properly it could have only dependency on the C library and maybe expat or xml since it basically would only need to output formatted text (instead of handling objects directly, they are handled by the browser)

{with this in mind I have started a framework for busybox httpd cgi scripts to automatically generate a simple gui and run apps (similar to gtkdialog) - so far it works with basic command line tools already ... I have no security included yet so no public releases and running only on localhost}

[ttuuxxx]

About 4 months ago when I compiled gtk3 I was not impressed at all with it, due to lack of compatibility with gtk2, funny how sfs and gtk don't care about supporting previous works, eg. Gtk 1,2,3, well atleast some gtk2 apps will work on gtk 3, unlike gtk1 vs gtk2. Well this new spin you have on gtk3 looks promising on a ultra thin client, but how would it fit into the big picture? like Gimp and other larger gtk apps.
ttuuxxx

[technosaurus]

... but how would it fit into the big picture? like Gimp and other larger gtk apps.
ttuuxxx

I've been following the commits and mail list to see if the xcb branch would ever take off or if they would ever get to a full cairo backend (I tried compiling with no X11 and cairo-xcb -- gdkpixbuf build went fine but there were still X11 specific items in gtk) Unless there are _significant_ improvements to an api, I see no reason to use it when we have one that does work. Switching to cairo only or from x11 to xcb, I would consider significant, but mostly they have just replaced some rarely used gtk functions with some more commonly used functions from gnome and other utility libraries (incidently this may actually be a good thing for porting some gnomish things to puppy -- all speculative though)

Here are some of the things I can see with broadway:
Good stuff
1. remote operation of your own apps
2. cloud based hosting of _real_ applications on the cheap (just patch and recompile existing apps ...like gimp, abiword, gnumeric,...)
3. X-free (not Xfree) instant on desktops using webkit on the framebuffer (~1/2 second startup)

Bad stuff
1. Things that may need to be patched may include the file dialogs (unless it uses the <form><input type=file.... type handling that is built into the browser) ... the audio and video should be handled by html5, but other things that use local devices won't work (anything that uses /dev /proc etc... to stat data)
2. Patch fest awaits us anyways for changes to or removing deprecated APIs as part of the 3.X transition
3. Running a full gtk app on a server to interface with a browser would take a lot more resources than a simple static frontpage (or perl, php etc..) I'm hoping I can make my toolkit follow the gtk look to supplement for gtkdialog style functionality (I've almost gotten a basic working version of my woo-ff media converter going)