`XML#nodes(String)` Method is Extremely Slow on Large Files

[volodya-lombrozo]

I have one large XML file and run the following java code on it:

private static Collection<XML> findDefects(final XML report) {
  return report.nodes("/defects/defect");
}

It runs extremely long (minutes). Could you speed up this method, please?

Here is my XML:
Pointer.xmir.txt

XML is instance of XMLDocument.

[volodya-lombrozo]

@yegor256 Could you have a look, please?

[yegor256]

@volodya-lombrozo why do you think it may be faster?

[volodya-lombrozo]

@yegor256 I showed it here: objectionary/lints#210

[yegor256]

@volodya-lombrozo how can we fix here? Implement our own XPath machine? )

[volodya-lombrozo]

@yegor256 I haven't checked it yet.

[yegor256]

@volodya-lombrozo jcabi-xml is simply an adapter for XPath Factory from JDK. We don't make it any slower and we can't make it any faster.

[volodya-lombrozo]

@yegor256 I'm not sure here. For now, it seems that you are right. However, I took a fast look on the code and have found that XMLDocument has tough synchronisations that might kill benefits of using parallel execution:

synchronized (this.cache) {
    return (T) xpath.evaluate(query, this.cache, qname);
}

This issue requires at least some investigation.

[volodya-lombrozo]

btw 81% of time we usually spend in this part of the code I showed above.

[yegor256]

@volodya-lombrozo this one may be closed, I believe, in favor of https://github.com/volodya-lombrozo/xnav

[volodya-lombrozo]

@yegor256 Yes, thank you!