Re: [xml-dev] XML Performance in a Transacation

[Tatu Saloranta <cowtowncoder@yahoo.com>]

--- Rick Jelliffe <rjelliffe@allette.com.au> wrote:

> Tatu Saloranta said:
> > --- Rick Jelliffe <rjelliffe@allette.com.au>
> wrote:
> >
> > Huh? So is this not a basic common knowledge?!? Of
> > course proper reusing of components can have
...
> It is one thing to be common knowledge, it is
> another thing
> to be common practise :-)

True.

> > It sounds more like developer education issue,
> though.
> > An order of magnitude or two simpler than trying
> to
> > hand-code assembler level ultra-efficient decoding
> > (although, for much bigger audience -- only small
> > number of people need to write libs, compared
> hordes
> > of developers using them).
> 
> I don't think it is an education issue. People
> implement as well as they can, given limited
resources. Last time
...
> So the solution is not sneering, but contribution.

Definitely. But what I meant by education issue wasn't
labelling it as "clues for clueless", but making the
relevant easy basic rules more accessible (writing
articles, blogs, tutorials, whatever).
And I assume knowledge (about tradeoffs involved with
usage patterns) is not a permanently limited resource.
That is, with more knowledge you can choose more
efficient approaches without using more effort.

Some things to know may be involved, but there are
many simple straight-forward things that yield very
significant benefits, compared to naive
implementation.
For example, reusing thread-safe factory objects in
stax (or sax, dom, to a degree) generally has
significant impact (from sizable to drastic), and is
easy to do.

Or, regarding XSL, even basic concepts: which things
are of linear complexity, and what are features that
are very likely to be expensive. The downside (wrt
performance) of functional languges is that complexity
involved is less transparent than with procedural
approaches: it is easy to use an approach that is
elegant and compact yet dead slow, compared to
something that is more verbose but efficient.
There's nothing new there I guess; new CS students are
taught recursion in all the wrong places, and
generally learn quickly enough not to calculate
multiplication by recursive addition by one. Same
should be possible with more advanced languages.

-+ Tatu +-



__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com