Re: [xml-dev] Features of XML Languages that Increase Complexity?

[Frank Manola <fmanola@acm.org>]

Roger--

I think you need to caveat some of your statements, such as: 

> Clearly if a feature elevates the language's complexity to "recursively enumerable" then you will want to avoid that feature.

and 

> I reckon there's not much point in creating an awesome XML language if its complexity exposes input-processing applications to widespread vulnerabilities.

The Dartmouth folks (very interesting work BTW) were very careful to talk about eliminating *unnecessary* complexity, and providing only the *minimum* power to do the job.  They seem to recognize, as some of your statements don't, that in order to do the job, sometimes the language must have a given level of complexity, and you have to decide to either do the job and accept (but try to mitigate to the extent possible) the vulnerability, or not get the job done.  After all, people built walls around castles for added security, but they still put doors in them.  Sometimes you have to get in and out!

I have no problem with:

> Science to engineers: some problems are not solvable, do not set yourself up to solve them. 

but at the same time:  

Engineers to science:  some problems are not completely solvable, but sometimes we have to do the best we can anyway.


--Frank