Article:		Binary XML, Again
Subject:		Too many words
Date:		2003-08-14 12:39:37
From:		Kendall Clark
Response to: Too many words
I'm not sure what counts as an "idea" for you, but I'll try to do better next time. (And if you think this piece is wordy, you should have seen it before Edd cut it down a bit. So I'm probably guilty as charged there, too.) However, I fail to see what is "ridiculous" about asking for empirical data to back up performance claims about binary variants of XML. Frankly, I think that's just the bare minimum of doing good research and engineering. As for whether I "demanded" data, I'll let other, perhaps more charitable readers decide. (Which, of course, all begs the question: so what if I *did* "demand" data...Performance claims without some empirical grounds are pretty meaningless, as just about everyone outside the Marketing Department concedes.)

Previous Message

Up

Next Message

• Too many words (Kendall has a hammer of his own)
  2003-08-15 13:01:23 jonnie savell [Reply]
  
  Rather than defining idea, let us define fluff (the absence of idea): "I dissent from several points of XML Orthodoxy because I am by nature, personal inclination, and experience, a dissident ..." Several spots of fluff I think. Too many words I think (and I fear that I have added too many of my own).

  
  OK, I got busted for the having made the charge that the demand was ridiculous. "Frankly, I think that's just the bare minimum of doing good research and engineering." Right you are. I was hoping that you would do some of it. I will back off because I have to some good research of my own.

  
  I get this eery feeling that this discussion risks missing a lot of territory if it continues to focus exclusively on document wide compression schemes.

  
  I like the xsi:type attribute. I think that the schema folks ought to add some built-in primitive types to their collection. I might like to strip from my application any "to string" conversion because I just don't enjoy converting well defined data types to strings. I might not mind converting my language specific data type format to a language neutral data type format, however. Hey, look at me (and stop laughing):
  ...
  <balance xsi:type="xsd:big_int">... unreadable stuff ...</balance>
  ...

  
  I avoided converting my data types to strings. I would understand if you felt that I made too big a fuss about this. I too might like more. I would like to tell the validating parser "Hey, ignore the next 16K of data following my <big_data> element. Just make sure that there is a </big_data> tag following the 16K of stuff and that everything is valid after that.

  
  So, is this pure noise, or is there some signal?

  
  
  
  

  • Too many words (Kendall has a hammer of his own)
    2003-08-15 13:14:28 Kendall Clark [Reply]
    
    --Rather than defining idea, let us define fluff (the absence of idea)...Several spots of fluff I think. Too many words I think...---

    
    Gee, since you're seriously pursuing the question here, publicly, I will respond in kind. There are, of course, some sentences of setup and teardown in every good column, and while you seem not to enjoy my way of doing that, it isn't fluff or misleading. If you were to try writing a weekly column, you might find that a certain kind of easy, personal rapport with one's audience is a good thing.

    
    --"Frankly, I think that's just the bare minimum of doing good research and engineering." Right you are. I was hoping that you would do some of it.--

    
    I'm sorry, but that's just an absurd requirement of a weekly column, of any kind, much less of the XML-Deviant. Perhaps you don't realize but the XML-Deviant column is supposed to survey and report on conversations of general interest in and to the XML developer community.

    
    Second, and more crucially, the obligation to back up performance claims with data rests with the person who's making the claims. I am *not* one of the people who's agitating for the W3C to define a general compression scheme or a domain-specific one. I don't have any obligation to do research and engineering around an approach I don't find interesting or useful. (Further, I might add, except for pretty rare exceptions, this kind of work is typically done in a funded environment (university, R&D lab, venture-cap backed startup...), *not* by technical press columnists.)

    
    --So, is this pure noise, or is there some signal?--

    
    So, you should attend the workshop, the announcement (and some of the issues) of which I reported on in my column. I don't find this to be a pressing issue; but then, as my respondents have informed, me I'm obviously not very bright! :>