Closed johnspackman closed 6 years ago
I wouldn't mind. In the end, qooxdoo-cli has no purpose of its own, it is fundamentally dependent on the compiler (even the contrib commands make no sense without it). So yes +1 for merging.
As long as the compiler api stays public i agree to the merging. We could handle the naming problem this way:
We should think about removing some of the older qooxdoo repositories. But that is another story.
👍 i'll get started on that during the holidays hopefully.
see here too: https://github.com/qooxdoo/qooxdoo/issues/9471
done with qooxdoo/qooxdoo-compiler#69
An ongoing difficulty with cli and compiler is that they are very closely linked (almost co-dependent) and yet have independent development paths; this leads to situations where the travis/sausage unit tests for one can fail because one of the projects is not using the correct branch of the other project.
Making Travis tests depend on published npm versions isn't suitable because
compiler
is required to go through it's own unit test cycle before merging into master and being published to npm; but some of thecompiler
unit tests rely oncli
to get built, and if there is a change incompiler
which breakscli
, then this forces us to release something to npm that breaks the livecli
until it has gone through it's own test/publish cycle.It seems that we would sometimes (often?) want to tie branches of
cli
andcompiler
together, have them unit tested together, and released together.How would it be if we merged
cli
andcompiler
?? Both are APIs so there shouldn't be much conflict, and at present the three contributors are the same people so it seems like a natural fit from a time-management point of view; fundamentally it would make releasing so much simpler because the unit tests would be self contained.