Sylvain Wallez wrote:
Carsten Ziegeler wrote:
Sylvain Wallez wrote:
This usage in CForms has already been introduced by the recent
library stuff, which associates prefixes to libraries, thus
effectively forbidding the use of ":" in widget ids (otherwise you
cannot differenciate between a widget name and a composite name that
references a library widget).
That is why I chose this character. The "/" and "." are also
forbidden (used for lookup paths). The "." cannot be used as it is
used to combine widget names in the generated IDs, and thus would
lead to a similar problem as the current one: "-" can conflict with
siblings, and "." can conflict with children.
Do we already validate a widget id if it does not contain all of these
forbidden characters? If not, we really should check this and throw an
exception when the model is read. Early failing is better than
unpredictable results later on.
Yep. The "." and "/" are already checked in
AbstractWidgetDefinition.setCommonProperties(). We just need to add ":".
BTW, I'm ready to commit the updated stylesheets, which I tested on IE
6, Firefox and Safari.
Why we need to use a symbol at any cost ? Can we use a simple word
prefix? As cform-[widgetID]?
I was not on GT2005, but I hear a lot about conventions was there. I
think this is good. Then let's define a fixed prefix name for cforms
names, declare it as a "cocoon keyword".
KISS rules! ;-)
WDYT?
Best Regards,
Antonio Gallardo.
