On 5/3/24 12:38, Pedro Falcato wrote:
On Thu, May 2, 2024 at 7:17 PM Kinney, Michael D
<michael.d.kin...@intel.com> wrote:
-----Original Message-----
From: r...@edk2.groups.io <r...@edk2.groups.io> On Behalf Of Pedro Falcato
Sent: Thursday, May 2, 2024 10:51 AM
To: devel@edk2.groups.io; Kinney, Michael D <michael.d.kin...@intel.com>
Cc: r...@edk2.groups.io; Leif Lindholm <l...@nuviainc.com>; Andrew Fish
(af...@apple.com) <af...@apple.com>
Subject: Re: [edk2-rfc] [edk2-devel] Proposal to switch TianoCore Code
Review from email to GitHub Pull Requests on 5-24-2024
On Wed, May 1, 2024 at 6:44 PM Michael D Kinney via groups.io
<michael.d.kinney=intel....@groups.io> wrote:
<snip>
* All contributors, maintainers, and reviewers must have GitHub IDs.
* The commit message would no longer require Cc:, Reviewed-by:, Acked-
by:
or Tested-by: tags. The only required tag would be Signed-off-by.
I'd just like to note that losing the CC:, Reviewed-by:, etc is a big
loss. Gerrit auto-adds Rb's, github PR's do not (I'd guess there's a
way to pull that off with github actions, but I haven't looked). It'll
be a mess if I have to go through online GH PR backlogs just to find
who to CC/add-to-review. It kills the decentralized bit off of git too
:)
Can you provide more details on the impact of the loss?
In my view, commits should be fairly self-describing. What changes,
why, are obvious, but who looked at it, who reviewed it, who was cc'd
but didn't respond, who tested are also pretty important. Git is
supposed to be decentralized, let's not forget. If we ever migrate
from GH, if GH ever goes down, if the links ever go down, you'll never
be able to know who looked at it. If you're looking at an EDK2 commit
deep into an Intel-internal fork, you won't know what "PR #478" is
(heck, rebase-and-merge doesn't reference PRs either).
Well said. That's my concern as well: TianoCore won't use GitHub
forever, and any GitHub metadata (PR numbers, GitHub IDs, bug numbers,
etc.) will become meaningless once we change. Never mind that the code
can be disassociated from the metadata simply by forking to a new
repository, as Pedro said....
Side-note: How are we supposed to find the PR for a given commit?
Searching doesn't seem to work well. For instance, I picked a random
non-trivial commit out of the current open PRs:
MdeModulePkg/Bus/Spi/SpiBus: Adding SpiBus Drivers.
https://github.com/tianocore/edk2/pulls?q=is%3Apr+is%3Aopen+MdeModulePkg%2FBus%2FSpi%2FSpiBus%3A+Adding+SpiBus+Drivers
has no matches?
I am curious how other GitHub projects handle this topic. I see it
I don't think they do, sadly. But I also don't know many people with a
positive opinion on GH PRs :)
Yeah... my opinions are decidedly mixed. They are convenient, but have
some serious gaps around archiving, auditing, and versioning of review
requests. They don't even let you review the commit messages (one of
their most serious flaws!)
<snip>
It is sad that we're moving to PRs after I finally got a nice and
sane(ish!) email workflow (openfw.io + b4). Otherwise, no objections,
it's better than edk2.git's half-email half-PR frankenprocess.
I'd guess this change only encompasses edk2.git? How about the other
repos? Any timeline for those?
The plan is to apply this to all repos, one at a time. Need to get the
revised process documented and working in one repo before applying to all.
Gotcha, thanks!
--
Brian J. Johnson
Hewlett Packard Enterprise
-=-=-=-=-=-=-=-=-=-=-=-
Groups.io Links: You receive all messages sent to this group.
View/Reply Online (#118826): https://edk2.groups.io/g/devel/message/118826
Mute This Topic: https://groups.io/mt/105873467/21656
Group Owner: devel+ow...@edk2.groups.io
Unsubscribe: https://edk2.groups.io/g/devel/unsub [arch...@mail-archive.com]
-=-=-=-=-=-=-=-=-=-=-=-
