definition for PRERELEASE and BUILD doesn't conform to semver 2.0.0 about semver-tool HOT 6 CLOSED

go ahead with c), it would be nice if you could provide a matrix of what is/was accepted/rejected.

from semver-tool.

I just wanted to edit the source, as i discovered that the regex already accepts dots as parts of PRERELEASE and BUILD. Maybe the patch will reduce itself to just change the wording of the documentation.
stay tuned.

from semver-tool.

I think the patch attached is to the best of my abilities (sadly, i must say).

from semver-tool.

To summarize: Indeed i couldn't find any combination that didn't pass unit testing.

I wasn't convinced that my idea to handle error-cases in unit-tests is the best, so i didn't include it.
My idea would be this: reserve some special return value that's impossible normally. For example the value "@error". Write this value into the README.md in place where the expected result would be written. Then the script test/documentation-test could be extended to detect this value and then expect a non-zero exit status from semver.

Regarding the specified comparison rules of semver2.x with regards to prerel, i am not confident enough to say there's a bug without testing. but i didn't want to test without having the tests documented and repeatable -- thus the above mentioned feature of test/documentation-test would need to be resolved first.

However, i must be frank, that i could not guarantee time to go through with all the steps necessary to change test/documentation-test. therefore i will only promise to go through with the changes as proposed (and discuss any problems you might find).

from semver-tool.

As I had already opened the code box to fix issue #33, I decided to jump in and work on semver spec 2.0.0.
There is already an open pull request (#30) addressing this issue, so I don't want to yank this contribution around. Herr Pogner ( @qyanu), if you have moved forward on the pull request or plan to shortly, let us know.
On the other hand, if your PR is still at the 1-Aug-2019 state and is not going to more forward (you hinted above about time constraints and test-case doubts), I'm ready to publish a PR that should close this bug.

1. It complies with 2.0.0.
2. Automated unit tests show previous non-compliance as well as testing compliance. Basis for the "matrix of what is/was accepted/rejected" requested by @fsaintjacques.
3. No changes to test/documentation-test.
4. No CLI changes.
   I'll be publishing my work shortly and will send a PR unless you'd rather I don't.

from semver-tool.

@ranger6 no no, i don't even read emails daily anymore. good days to you!

from semver-tool.