| 1 |
On Tue, 27 Nov 2018 15:10:36 -0500 |
| 2 |
Rich Freeman <rich0@g.o> wrote: |
| 3 |
|
| 4 |
> Also, GLEP 76 as it is written says: "Projects using this scheme must |
| 5 |
> track authorship in a VCS, unless they list all authors of |
| 6 |
> copyrightable contributions in an AUTHORS file." |
| 7 |
|
| 8 |
Idea: How about using VCS as a defacto set of AUTHORS, but *also* support |
| 9 |
an AUTHORS file that is designed to extend the content that git |
| 10 |
provides generically. |
| 11 |
|
| 12 |
That way you can just say something like: |
| 13 |
|
| 14 |
"if the name appears naturally in git shortlog, you don't need to add |
| 15 |
anything to the AUTHORS file" |
| 16 |
|
| 17 |
And then git2rsync conversion can unify the two input sources. |
| 18 |
|
| 19 |
nb: generating the AUTHORS file from git is naturally very time |
| 20 |
consuming, as it requires full traversal of the entire repository. |
| 21 |
|
| 22 |
However, there are practical ways of caching this (eg: generate it, |
| 23 |
record the SHA1 it was generated at, then, next time, simply traverse |
| 24 |
the subrange between HEAD and SHA1-last and update the cache based on |
| 25 |
that). |
| 26 |
|
| 27 |
But that very much puts this in the realm of "things that are painful |
| 28 |
for end consumers to actually do themselves" |
Archives