search

rillig / gobco

Measure branch coverage of golang tests
62 stars 12 forks source link

master branch broken and no tagged release versions #3

Closed KantarBruceAdams closed 5 years ago

KantarBruceAdams commented 5 years ago

Hi, The current head commit doesn't work (for me anyway) as follows:


go get github.com/rillig/gobco
go: downloading github.com/rillig/gobco v0.0.0-20190712181434-65bb23983152
go: downloading github.com/google/uuid v1.1.1
# github.com/rillig/gobco
/main.go:82:16: undefined: os.UserHomeDir
../../../go/pkg/mod/github.com/rillig/gobco@v0.0.0-20190712181434-65bb23983152/main.go:204:24: exitErr.ExitCode undefined (type *exec.ExitError has no field or method ExitCode)

Apologies if this is teaching your grandmother to suck eggs but I would recommend having separate feature branches for experimental changes and keeping only tagged release versions on the master branch. See for example, https://datasift.github.io/gitflow/IntroducingGitFlow.html

Regards,

Bruce.

KantarBruceAdams commented 5 years ago

I would suggest tagging commit ff234e5b7147368487f0012a994e5e1f12cdd74a as 1.0.0 (or maybe 0.1.0) as it predates the latest tinkering.

rillig commented 5 years ago

Regarding your suggestion of adding feature branches: I won't do this right now because I always think of the master branch as being as stable and useful as possible. Granted, before copying everything to a temporary directory, gobco wasn't even ready to be used reliably, but feature branches wouldn't have changed this.

Feature branches also wouldn't have caught this use of new API. Therefore the argument of adding feature branches is moot in this particular case. It worked properly for me, I had tested it, therefore I considered it stable enough to be published.