>--------------
>However, this whole game did have a purpose, even though it may not have been clear in the outset :). We've found a few good ways to skin the approximate virtual cat, tossed ideas around and had good time. The initial C# vs VFP comparison was just the nail for the soup.
>[explanation of story here]
>---------------
>
>
Dragan, you don't know me very well. My university degree is in history and literature, with a great deal of time devoted to anthropology and folklore ...
Hey, what do we have conferences for? We could just bore everyone to... the point of collective snore :).
>Anyway, FYI, the English variant of that story is called "Stone Soup". You might be interested to see how often the metaphor is used by programmers; in fact I think there is an Open Source contribution-community that goes by that name <s>.
I once wanted to write "Programming in 100 jokes", but wrote down only eight, and those were only the ones which translated well.
And I wanted to translate "nail" into "stone", because my intuition told me so (i.e. I probably heard the English version sometimes), and wanted to avoid the association to fingernails (English is so OOP, most of the operators, aka words, are overloaded already), but hey, it's a Sunday, so I went the folklore way.
>As I said before, I am glad that everybody had a good time and many good performance-enhancing ideas were brought up. I don't mean to denigrate the value of the discussion or of the exchange of ideas.
>
>However, the comparisons of relative times of the various methods were fallacious -- and this has nothing to do with XML in particular.
>
>Each method would have had additional function calls -- and I believe that, because each method structured its code differently, you might have seen different relative times if the proper function calls had been added.
>
>IOW "Well-formed XML" and its requirements (the additional calls) becomes a real-life illustration of why these comparisons don't really matter very much and final conclusions should not be drawn from them.
Right, that's why I called it fake-XML. And building a fake xml was a good enough approximation of a real-life task, nothing more.
>-----------------
>I know I'd never try to write real XML walking (umm... doesn't translate well, take it as "in a pedestrian way").
>--------------
>
>Translates fine, Dragan (and incidentally is reminiscent of the French "ca ne marche pas", which I have always liked <g>).
>
>Never say never. As I said, I often do "raw writes" -- there are different reasons in different applications. I test performance of all possibilities with the application scenario in mind, before settling.
Of course. There was a time (probably when LLFF were introduced) when I was solving a bunch of real life problem with raw reads and writes... and never forgot what can be done with them. It's just a matter of estimating the importance, frequency of use, time to develop, time to test etc - and if it shows that raw write is the way to go, naturally I'll go that way again.