Re: Second MNG-6964 and MNG-6967

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view
|

Re: Second MNG-6964 and MNG-6967

rfscholte
LGTM
On 24-7-2020 08:47:47, Dennis Lundberg <[hidden email]> wrote:
Hi,

First up, I'm sorry. I made a commit to maven core on gitbox without
requesting a review first. Thanks to Sylwester for pointing that out. So
here comes a belated request for review.

MNG-6964 describes an edge case where rather unusual version numbers are
not sorted correctly. After investigating the code in maven-artifact I
found that to be true. If a qualifier (something that comes after a dash
'-') starts with "0." and it was being compared to a version without a
qualifier it was deemed to be equal. I changed so that instead of just
checking the first component of the qualifier it checks all of them looking
for any difference. Tests were added and all existing tests pass as well.

https://issues.apache.org/jira/browse/MNG-6964
https://github.com/apache/maven/commit/4f193b3fc26c2ccf2a1b7a34917faccedac1ea0e


When investigating this I found that the command line output from
maven-artifact could be a bit clearer. Also it doesn't match exactly what
is written in the documentation, leading people (myself included) to draw
the wrong conclusions. In MNG-6967 I describe this. The change is purely
cosmetic, but still important.

https://issues.apache.org/jira/browse/MNG-6967
https://github.com/apache/maven/commit/51c0399030dcb0b88080a7f9e50c3d09c6913dda


Thanks,
Dennis Lundberg