Re: [xml] libxml2 performance

["Nix N. Nix" <nix go-nix ca>]

On Mon, 2004-12-06 at 12:53 -0500, Rich Salz wrote:
> I think you missed Daniel's point.  You can find out how long it would 
> take to parse the data just by running his xmllint command.  That's all 
> you need to know, right now, is the parsing effort.

I know that. Unfortunately, I don't have any data in XML format.

> I don't know what your data currently looks like

The smallest useful example:

[VERSION]
  qcadesigner_version=2.000000
[#VERSION]
[TYPE:DESIGN]
  [TYPE:LAYER]
    type=1
    status=0
    pszDescription=Cell Layer
    [TYPE:QCADCell]
      [TYPE:QCADDesignObject]
        x=135.000000
        y=176.000000
        bSelected=FALSE
        clr.red=0
        clr.green=65535
        clr.blue=0
        bounding_box.xWorld=126.000000
        bounding_box.yWorld=167.000000
        bounding_box.cxWorld=18.000000
        bounding_box.cyWorld=18.000000
      [#TYPE:QCADDesignObject]
      cell_options.cxCell=18.000000
      cell_options.cyCell=18.000000
      cell_options.dot_diameter=5.000000
      cell_options.clock=0
      cell_options.mode=0
      cell_function=0
      number_of_dots=4
      [TYPE:CELL_DOT]
        x=139.500000
        y=171.500000
        diameter=5.000000
        charge=8.010000e-20
        spin=0.000000
        potential=0.000000
      [#TYPE:CELL_DOT]
      [TYPE:CELL_DOT]
        x=139.500000
        y=180.500000
        diameter=5.000000
        charge=8.010000e-20
        spin=0.000000
        potential=0.000000
      [#TYPE:CELL_DOT]
      [TYPE:CELL_DOT]
        x=130.500000
        y=180.500000
        diameter=5.000000
        charge=8.010000e-20
        spin=0.000000
        potential=0.000000
      [#TYPE:CELL_DOT]
      [TYPE:CELL_DOT]
        x=130.500000
        y=171.500000
        diameter=5.000000
        charge=8.010000e-20
        spin=0.000000
        potential=0.000000
      [#TYPE:CELL_DOT]
    [#TYPE:QCADCell]
  [#TYPE:LAYER]
[#TYPE:DESIGN]

I added indentation for clarity.

> , but you could probably 
> build a  (small?) testcase of your expected input format without a lot 
> of work.

I'm working on converting the above to xml that complies with the Schema
I mentioned earlier. I will then copy the single QCADCell object many
times over.

Meanwhile, I would like to know if anybody here knows of instances when
libxml was asked to parse a file that's approximately, say, 45MB, and
how long it took to parse it. I just need to know if libxml performs
significantly better than 39s/23MB of text.