Don't include 0.x updates in --target minor

[danwdart]

• I have searched for similar issues
  (sorry if I missed it, tried to look for 0.)

I don't think we should include 0.x updates in --target minor. It's not quite the same as --target semver because if the user specified bounds of only 0.x.y and not 0.x.z, --target minor should update to 0.x.z anyway, and not to e.g. 0.w.0.

---

Steps to Reproduce

1. Have a package.json pointing to e.g. axios 0.18.x
2. Run ncu --target minor.
3. Observe that axios has been updated despite semver stating here that 0.x is inherently unstable and that changing x in 0.x.y is not necessarily a minor change, only after v1.

Current Behavior

0.x versions are updated using --target minor.

Expected Behavior

0.x versions to be left alone using --target minor.

Thanks!

[raineorshine]

Previous discussions in #958 and #1213.

After having a couple of these discussions, I've ended up here:

• I'm not comfortable with --target minor including v0.x, as it leads to confusion.
• I'm not comfortable with --target minor excluding v0.x, as it is too close to --target semver and will lead to a different kind of confusion.

So my tendency is to leave things as they are, since I'm not really satisfied with either behavior. I'm also leaning towards leaving things as they are since there are several workarounds:

• Exclude v0.x using filterVersion.
• Exclude v0.x with a custom filterVersion function.
• Use --target semver (is is slightly different as you pointed out, but probably perfectly fine to use in most cases).