I'm not sure how to resolve this. We use "warn" routinely inside
problems to
help write and debug problems. pretty_print_rh formats complicated
structures in tables. This is very useful and I'd rather not loose it.
Hard coded warn messages can have < > manually replaced by < and
> but
there is a problem if the warn message has a variable.
I suppose that the long term solution is to build a protected print or
debugging statement
within the PG environment that allows us to produce debugging output
without using the warm
mechanism.
For now I'd like to leave it the way it is, with HTML allowed in warn
messages, and you have to
do your own escape. If this causes too many problems I'll up the
priority of finding another method
for reporting debugging errors in problems.
Take care,
Mike
On Jan 15, 2005, at 2:51 PM, Samuel Hathaway wrote:
> On Sat, 15 Jan 2005, Mike Gage via activitymail wrote:
>
>> Commented out escapeHTML() of warning messages since it prevented
>> using tables when outputing info about answer evaluators.
>>
>> Can someone tell me whether there is another reason why not allowing
>> HTML in warning messages is or would be important? If so we can
>> figure out another way to debug answer evaluators.
>
> The intent is not to prohibit HTML in warning messages but to allow the
> literal display of characters like '<' and '&' in warnings.
>
> It's going to be a problem if we need to pass HTML through the warning
> facility, since there would be no way for the warning handler to
> differentiate between a warning that just happened to use characters
> like
> '<' and '&' (say, generated by Perl), and a warning that has purposeful
> HTML in it that ought to be passed through unmolested.
> -sam
>
|
