MASSEMBLY-945 - speed improvements

[gnodet]

This PR only makes sense with
codehaus-plexus/plexus-utils#106
codehaus-plexus/plexus-io#33
codehaus-plexus/plexus-archiver#157
apache/maven-common-artifact-filters#15

... but if I add the snapshot upgrades, the checks won't pass anymore ;-)

[elharo]

have you measured the improvements? if so, what are the numbers?

[bmarwell]

have you measured the improvements? if so, what are the numbers?

Null-length arrays are known to be faster since at least 2013 iirc.

• https://shipilev.net/blog/2016/arrays-wisdom-ancients/ (updated 2016)
• https://stackoverflow.com/a/29444594/1549977 (from 2013)

In my opinion measurements are not needed here because of this.

[elharo]

Far more often than not, microoptimizations like this result in no measurable impact. All claims of performance improvements need numbers.

[bmarwell]

Far more often than not, microoptimizations like this result in no measurable impact. All claims of performance improvements need numbers.

codehaus-plexus/plexus-io#33 (comment)

[gnodet]

The speed gain may be negligible for this PR, it does not matter much.
However, I'd like the other PR referenced above to get merged, so that we can upgrade to those new versions to have a non negligible performance gain.

[eolivelli]

+1