Duplicate headers in the changelog

[jan-molak]

Hey @jpnelson!

I noticed that if a given package has not been changed as part of the commit, its changelog contains a duplicated header of the latest version, but with the current date:

<a name="1.0.1"></a>           <- no change, just another build,
## 1.0.1 (2017-04-13)             yet the header is here



<a name="1.0.1"></a>           <- a change happened here, OK
## 1.0.1 (2017-04-01)

I'm using the independent versioning mode.

[lorenzomigliorero]

+1

[lorenzomigliorero]

I @jpnelson , thanks for your work.
I encounter the same bug, headers are duplicated.

These are my last two commits message:

chore(release): releasing component

affects: @malmo/cli@1.0.3

Released from sha 336b1214ac017521c8ea3c235cf89910e2f51811

[skip ci]
Tag for lerna release

fix(yarn): fix module resolution of alias dependencies

affects: @malmo/cli

The first one is auto-generated from lerna-semantic-release pre command.
After launch lerna-semantic-release post this is the content of the generated @malmo/cli CHANGELOG.md:

<a name="1.0.3"></a> // this row is generated by a not relevant commit (releasing component). WRONG
## 1.0.3 (2018-11-14)



<a name="1.0.3"></a> // this row is generated by a relevant commit. RIGHT
## 1.0.3 (2018-11-14)

After investigating, this is the responsible line:

lerna-semantic-release/packages/lerna-semantic-release-analyze-commits/index.js

Line 28 in 1a9b055

return thisPackage.substring(0, thisPackage.lastIndexOf('@')) === packageName;

What's the purpose of this row, that return a false positive when thisPackage is equal ti @malmo/cli@1.0.3?
I think that this control isn't necessary, but i i'm not aware of all the possible edge case.

Alternatively, replacing this line

lerna-semantic-release/packages/lerna-semantic-release-analyze-commits/index.js

Line 39 in 1a9b055

return trimmedPackages.split(', ');

with this one return trimmedPackages.split(', ').filter(pkg => pkg.match(/@/g).length < 2);
it's enough, filtering by the @ occurrences exclude the false positive.