| Thread Previous • Date Previous • Date Next • Thread Next |
On 14/05/2020 18:43, Wayne Stambaugh wrote:
I am against rejecting any library contributions because of git to be honest. Most of our contributors are complete novices with git. I kind of supect they are mostly electrical engineers not software engineers so quite unlikely to need git knowledge in their career. We might already filter out a lot of potentially good contributors with the very small git knowledge that we need to require and i don't want to increase the negative impact of git.Rene, I wonder if we should implement a library commit message policy similar to the one we use for the source repo[1] and tailor the changelog tags for how library commits (or any other kicad repo) are made? It's pretty painless from a committers point of view since you already have to provide a commit message with git. Maybe a "PACKAGING" tag that could easily be extracted would be helpful. It probably wouldn't hurt to have this tag in the source repo as well so we don't blindside package devs with new dependencies and folder layout changes. Cheers, Wayne [1]: https://docs.kicad-pcb.org/doxygen/commit_messages.html .
Right now we librarians write new messages for our contributors during the merge process (we use githubs squash feature). Because of that we can not use CI and not even a manual review to ensure following of some commit message rules. A fully manual workflow is extremely error prone so i doubt we will ever get a full changelog from git.
To be honest something similar to our changelog script that analyzes symbols might in the end be a better fit for the whole lib. Even better would be a graphical diff tool to be honest. And all of that build into kicad itself for example as part of the "update footprint/update symbol" tools. After all the library assets are graphical things. Describing changes to them using text is really not helpful at all.
| Thread Previous • Date Previous • Date Next • Thread Next |
