Re: Dynamic errors, choose

Richard Tobin wrote:
> End of section 3:
> 
>   "If an XProc processor can determine statically that a dynamic error
>    will always occur, it may report that error statically."
> 
> Does this mean it may refuse to run the pipeline, or merely that it
> may report the error?  What if the error would have been caught by
> a try/catch?

I think that it MAY report the error and therefore MAY refuse to run the 
pipeline (that's what you usually do when you hit a static error).

If the error would have been caught by a try/catch then the XProc 
processor is non-conformant. It's only allowed to report the error 
statically if it can determine statically that a dynamic error will 
always occur: if it would have been caught by a try/catch then it 
wouldn't always occur, so it's not the kind of dynamic error that an 
implementation can report statically.

> The example that brought this up:
> 
> As far as I can see, a p:choose with no when or otherwise is always a
> dynamic error, since no branch will be chosen.  It seems reasonable to
> report this statically and reject the pipeline.

Err, yes. I think the syntax of <p:choose> should reflect this (in other 
words, that <p:choose> *must* have at least one <p:when>, and I think it 
should probably always have a <p:otherwise> as well).

Cheers,

Jeni
-- 
Jeni Tennison
http://www.jenitennison.com