Re: [DotGNU]Xml Alien Invasion - A standards war

[Gopal V]

If memory serves me right, minddog wrote:
> address@hidden tests]$ ilrun a.out.exe 
> attr="foo">data</tag>
> address@hidden tests]$ mono a.out.exe 
> data</tag>
> address@hidden tests]
> 
> And .NET outputs Nothing ....

Actually they might be wrapping our reader with a further buffered reader...
though why they would want to do that is beyond me :)

> my opinion all these MoveTo* methods are redundant in the ecma standard, but 
> it comes down to being ecma compliant at the end of the day.  It will only 
> cost us an internal class and another wrapper method inside XmlTextReader.  

Could someone test if MoveTo* allows you to move back on attributes ? ...
Because if it can, then XmlTextReader is *not* a constant memory reader.

> What are your thoughts, should we lead the standard and fix redundancy in the 
> ECMA spec? I could help out with the generator and make it work with 
> XmlDocument since its the same across all implementations AFAIK.  

I wonder if we should write test cases and run them on all platforms ? ...

> That could save us from hurting speed we get from Reader at present.  

Is our reader *faaaast* ? ... From your own comments it's about 4x slower
than libxml2 (which might be good performance for a bytecode parser on an
interpreter ;) ... Yeah , we should keep it fast ... I wouldn't want a
parser that would take longer to parse than for me to read xml ;)

Gopal
-- 
The difference between insanity and genius is measured by success