Antranig Basman antranig at caret.cam.ac.uk
Sat Jun 23 18:20:25 UTC 2007

The firefox explosion was indeed finally tracked down to the Dojo
I18N libraries (by the painful and traditional process of binary
This line in the component startup was causing the Firefox event loop
to lock:

dojo.requireLocalization("fluid", "instructions");

As I mentioned in the previous mail, I don't think this is the 
right approach to I18N for our components anyway, so in this
commit I have destroyed the nls directory (sorry, Jonathan!) 
and commented out the parts of the tests which referred to it.
I have also "arbitrarily" adjusted the number of nodes in the 
whitespace Utilities test so that it passes, but as I remarked
to Anastasia in chat yesterday (now lost locally since it crashed!)
I don't think this DOM-sensitive approach is the right one for
the widget either, and hope that this whitespace logic will
eventually go away entirely. We should really be operating a
"swappage" model wrt. our lightbox cells rather than a "slippage"

I have moved the I18N for the widget over to a server-based
approach using the already existing messaging file. This dumps
all the appropriate localised messages in a hidden div at the
bottom of the outer part of the page, whose ID is handed in as
another initialisation argument to the script block. There is 
a local version of this block within the lightbox template for
previewing and development purposes.

Some non-crucial refactoring work to the server package structure,
persistence is still a little way off.

At revision 10246.


More information about the fluid-work mailing list