Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [cross-project-issues-dev] The "new user -> ECA -> signed-off by" workflow is broken

I added an action item to discuss on AC call next week.

--
jesse mcconnell
jesse.mcconnell@xxxxxxxxx


On Thu, Mar 4, 2021 at 9:33 AM Ed Merks <ed.merks@xxxxxxxxx> wrote:

Folks,

I did repeatedly try to get this change, posting arguments such as this below (in which I did to remove some email addresses to not post them here).  I got Matthias Sohn to also back up the arguments.  But so far to no avail...

I can try again, but it seems pointless to repeat the same arguments so perhaps the AC, would like to make a case that I can bring forward yet again?

______________________________________________

Hi,

I'm not sure this specific topic is actually one that needs to be reviewed and considered by the IP Advisory Committee nor that it even requires some type of Board approval...

The committer community has request via https://bugs.eclipse.org/bugs/show_bug.cgi?id=558653 that the requirement for commits to contain a Signed-off-by tag be eliminated.

I.e., this part of the handbook needs to change to remove the "Signed-off-by" tag requirement:

  https://www.eclipse.org/projects/handbook/#resources-commit

The fundamental point is that when one looks at a commit like this one:

https://git.eclipse.org/c/oomph/org.eclipse.oomph.git/commit/?id=a0bf842d3539f2a34516ccd2b1b257950875db37

Which looks like this when the email addresses are not filtered out by the web view:

commit a0bf842d3539f2a34516ccd2b1b257950875db37
Author: Christoph Läubrich <...> 2020-07-05 14:10:03
Committer: Ed Merks <ed.merks@xxxxxxxxx> 2020-07-11 08:05:19
Parent: 6c2e79c17db44454e8b518ba1da8654d4f77490c ([Releng] Eliminate deprecation warnings new to Java 11.)
Child: 25df8caf7ba3841e3e9efd63d83b567771ebe61d ([Releng] Build against 4.16 to avoid surprising Java 11 BREEs.)
Branches: change/165847/2, master, origin/master

[494735] Eclipse Installer does not create .desktop file for the
menu - add Linux Desktop support

Change-Id: I145791e1ab63278fffca199fb3660ca017e62a00
Signed-off-by: Christoph Läubrich <...>

It's clear that the Author:

  Author: Christoph Läubrich <...>

and the Signed-off-by:

  Signed-off-by: Christoph Läubrich <...>

specify the same information.

As such, any checking that's done on Signed-off-by tag can be done on the Author tag instead and similarly any IP tracking done via Signed-off-by can be done via the Author tag instead.

I other words, the Signed-off-by tag is redundant. Given it represents one more hurdle that contributors often forget, this requirement should be eliminated.  After all, the Author tag is required in a commit so contributors necessarily must specify one. This is sufficient for all verification and tracking purposes.

In terms of implementation effort by the staff, Mikaël Barbero says the following:

The business logic implementing the signed-off-by is currently replicated in various systems (gerrit, github, gitlab). While there are efforts currently to develop a centralized commits validation service, this check has already required a couple of re-work of the Gerrit ECA validation service after some backward incompatible upgrades of Gerrit. Getting rid of the signed-off-by requirement would greatly simplify the validation logic and would stick to the strict validation of the ECA signature requirement.

Thanks,
Ed


On 04.03.2021 16:08, Jonah Graham wrote:
Hi Wim,

Thanks for sharing your methodology. I can't speak for the 1% result, but it does not surprise me. The Step 8 is the biggest PITA for contributors on the GitHub PR flow. Many (most?) PR contributors on github never rebase or amend their commits (for other projects). I have had to regularly walk people through the flow of rebasing and amending changes.

Jonah

~~~
Jonah Graham
Kichwa Coders
www.kichwacoders.com


On Thu, 4 Mar 2021 at 09:52, Wim Jongman <wim.jongman@xxxxxxxxx> wrote:

Hi,
We lose 99% of our casual contributors.

Can you share some insight on how you got this high number? 

I have done a scientific measurement. See below how the process works in practice [1] with drop-off percentages in each step of the process.
 
Cheers,

Wim


[1]
This is the original mail that I decided not to send because it was typed in frustration. Take this with a grain of salt; it might not fully accurate reflect the current process:

I again want to ask to remove the "Signed-off-by" requirement. It is not needed because the author and the email are already on the commit. If the project is on GitHub, why do users need to have an Eclipse account when they already have a GitHub account?

It is a PITA because, according to my calculation, this workflow scares away 99% of casual contributors. Here is a replay of the workflow (GitHub):
  1. People file a pull request.
  2. Eclipse checks say: You did not sign the ECA.
  3. 21% of the contributors are confused or don't care and drop off. They say: You merge it for me (which we can't do because it is not our code.)
  4. The rest goes to the website to sign the ECA.
  5. It turns out you need to create an Eclipse account first, so another 53% does not bother. Why is this? People have a Github account; can't we work with that?
  6. The ones left make another change in their PR because that is the only way to retrigger the checks
  7. Eclipse checks say "no signed-off by" footer in the commit message. This is not needed because all information is already in the commit. IT MAKES NO SENSE!!!
  8. So the contributor has to REDO ALL THE COMMITS IN THE PR unless they are some git wizard that can update each commit comment: 25% says "Nah."
1% left of the 77% we still have after step 3
_______________________________________________
cross-project-issues-dev mailing list
cross-project-issues-dev@xxxxxxxxxxx
To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/cross-project-issues-dev

_______________________________________________
cross-project-issues-dev mailing list
cross-project-issues-dev@xxxxxxxxxxx
To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/cross-project-issues-dev
_______________________________________________
cross-project-issues-dev mailing list
cross-project-issues-dev@xxxxxxxxxxx
To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/cross-project-issues-dev

Back to the top