On 2025-09-28 22:13, Arsen Arsenović wrote:
Dragan Simic <[email protected]> writes:
On 2025-09-28 19:53, Collin Funk wrote:
Arsen Arsenović <[email protected]> writes:
Note that we're considering using Sourceware-hosted Forgejo (which
Codeberg maintains and uses) for the GNU toolchain, with the
advantages
above (especially automation-related advantages) in mind. See
https://gcc.gnu.org/wiki/ForgeExperiment
Have any projects begun using this yet? I am a glibc committer, and
as
far as I can tell no one has begun using it [1].
Maybe gcc and/or binutils have started experimenting with it? I do
not
follow their development closely.
Collin
[1] Well maybe they do for the Web UI which is nicer than gitweb and
has
features like code search.
If I understood the forge experiment [2] correctly, it implements a
GitHub- style workflow that allows people to use pull requests, but
the end result are actually patches sent to the "old-school" mailing
lists?
This is the case for length of the experiment, the copy on the forge is
read-only in a sense, and all patches merged are specified to be sent
onto the mailing list, ergo people have to do it anyway currently.
The end goal, to my understanding, is to not duplicate work, and skip
the ML step, eventually.
I don't see that written anywhere, but maybe I'm wrong.
If so, I find such an approach nearly perfect, because it merges the
best of both worlds and frees the new contributotrs from the need to
set up git-send-email(1), for example, if they prefer not to.
That's pretty much the same as what GitGitGadget [3] already allows,
which Git uses as an option for submitting patches. I've always truly
loved GitGitGadget's presence as an option, because it keeps the
mailing lists as the primary workflow, while it removes the proverbial
entry barrier for the contributors who actually prefer to use pull
requests. To phrase it a bit differently, variety is the proverbial
spice of life.
I agree that GGG fixes the "new contributor" issue.
But, that's the less important part of this experiment IMO.
Well, GitGitGadget is an "experiment" that's been running successfully
for many years, in parallel to the mailing list.
