This Week on C9 from Blizzcon with Mike "Sampy" Sampson

Posted By: Dan Fernandez | Oct 13th @ 9:23 AM
On a "very special" episode of This Week on Channel 9 from Blizzcon, Dan is joined by Mike "Sampy" Sampson as they discuss the interesting developer stories of the week:

- WPF Effects Library ships on Codeplex (video demo)
- Stephen Walther - ASP.NET MVC 5-Part series on building a Family Videos site, via Alvin Ashcraft
- Microperformance benchmark - XMLDocument vs XElement, XElement performs 5-10x faster, via gOODiDEA.NET
- Live Search now in Facebook via Sarah Perez
- Web Platform installer ships, via Scott Hanselman
- Mono 2.0 is out, via Miguel de Icaza
- Video clip of Charles Torre interviewing Anders Hejlsberg on adding language constructs to C# for concurrency
- Video clip of Bill Hill whiteboarding how fonts include thousands of lines of assembly code to improve readability
- Touchless Multi-touch (video demo) open-source SDK on Codeplex, via TechCrunch
- Sampy's pick of the week, the Diablo III Wizard class
- Dan's pick of the week, the Blizzard Costume contest with the scary undead rogue girl
Tags: , , ,
Rating:
2
0
#Oct 14th @ 8:18 AM
vesuvius
vesuvius
Everyone has talent at twenty-five. The difficulty is to have it at fifty.
I'm afraid that Alvin Ashcraft's test is fatally flawed. Using XMLDocument, the most efficient way to stream information is using the FileStream and not MemoryStream library.

Linq's disadvantage is that you have to load all the data into memory before you can perform actions upon the data. Using XMLDocument properly you would copy the data node by node, this would trounce Linq 365 days a year.

Close but no cigar!
#Oct 14th @ 9:04 AM
Dan
Dan

[FYI - It's actually Anil RV's test, not Alvin Ashcraft's test]

If you've looked at the test though, it doesn't test streaming XML from a filestream, instead it tests building an in-memory XML document dynamically. That's not to take away from your point, it's just that the test is only for in-memory XML document creation.

It would be great if you could build your own micro-performance benchmark on loading XML from a filestream to see the performance advantages/disadvantages of XMLDocument vs XElement (and probably XMLTextReader which I would think should be the fastest).

Cheers,
-Dan

#Oct 14th @ 9:22 AM
vesuvius
vesuvius
Everyone has talent at twenty-five. The difficulty is to have it at fifty.
I stand corrected I meant XMLTextReader. One of the first commercial applications I wrote did just this, and I opted for the lower level API's because it was processing XML data via webservices so loading data was not an option (XML files with lots of orders)

I wish I had the time to test this at present, but the XML team have these blogs that some people may find useful. I must add if I was faced with doing any XML stuff today, I'd still head straight for the base class I/O libraries, as I don't find them that difficult now.
#Oct 15th @ 5:29 AM
alashcraft
alashcraft
Whew... I'm glad my link to Stephen was not fatally flawed.   Smiley

Thanks for the linkage, Dan! I'll be at PDC with the DZone.com booth crew. Stop by and say hi if you have a minute.