Bonjour Peter,

Le Fri, 24 Jul 2015 11:28:44 +0200, Peter Stuge <[email protected]> a
écrit :

> Hi,
> 
> Albert ARIBAUD wrote:
> > > > This doesn't sound very encouraging for the project in general.  :-(
> > > 
> > > It's essentially the contributor's task to make it effortless for
> > > maintainers to accept contributions.
> > 
> > Maybe not effortless but as easy as reasonable, certainly.
> 
> Yes effortless. Remember that maintainer time is a limited resource.

"Effortless" is an ideal.

> > Here, IIUC, the main factor is the time needed to analyze the patch(es).
> 
> Right, good review always takes time, but when a patch is less than
> ideal then review requires a lot *more* time from the maintainer -
> the limited resource. This is unacceptable, not because maintainers
> hate non-ideal contributions, but because it is impossible for
> maintainers to spend so much time on overhead.
> 
> 
> > OTOH, it should not be burdensome to contribute.
> 
> I think that's a little bit too naive. When someone wants to
> contribute something and do it well they have to accept quite a lot
> of responsibility if they want to maximize the chances of their
> contribution being included into the project. Over the years I've
> seen countless developers who can do this pretty effortlessly, and
> I've also seen many developers who can not get a simple patch into
> acceptable form even after ten rounds of feedback. Since maintainers
> are not able to go back and forth with contributors ad infinitum the
> responsibility and yes burden for contribution really lies with the
> contributor.
> 
> I agree with you that projects should not raise unneccessary
> barriers for contributions, here some projects are better, some worse.

Unnecessary barriers are my point, really -- and just like a
maintainer's time should not be wasted, so should not a contributor's.
After all, without the contributor, the maintainer would have to patch
to review and apply.

> > Ok, that's anecdotal, but still: I've tried to register to the OpenOCD
> > Gerrit, and gave up for now because it wants me to register through a
> > GitHub, Google, Launchpad, or Yahoo! account, or through Open ID, and
> > did not leave me any visible option to just register with the e-mail
> > and password of my choice. *That*, for instance, is a burden, as these
> > accounts are not related to, an not needed for, contributing to OpenOCD.
> 
> Gerrit is used by many different projects, it is a very valuable tool
> and contributors only need a single OpenID in order to contribute to
> all of those projects. I'm happy to create an OpenID account on my
> OpenID server for you if you want, please email me off-list in that
> case.

(see below)

> > IMO, the amount of unrelated-to-OpenOCD effort that a contributor needs
> > to endure in order to submit a patch should be limited to providing a
> > verified e-mail address to Gerrit.
> 
> Gerrit does not support that mode of operation. Feel free to send
> patches to the Gerrit project to extend it, so that it does.
> 
> But please think hard about how Gerrit has outsourced the entire
> authentication problem by relying on OpenID before you write
> something up. Please also write contribute such a patch to Gerrit
> only with strong operational experience from "verified email"
> schemes, so that your patch already tackles the important problems
> involved.

OpenID is only the *default* mode of authentication with Gerrit. LDAP
with a local server is another possible mode, that allows registering a
local account -- and I suspect that's what Motorola was doing back in
2012 when I still worked there and used Gerrit without having any
OpenID involved at all.

And I still consider that if authenticating me by the fact that I can
receive mail on the address I've given is good enough for U-Boot or
Linux, then it is good enough for just about any project. Granted, that
rules out single sign-on (unless you set up a shared LDAP server, I
imagine) but I personally don't care for SSO.

Oh well, then. OpenID is required for OpenOCD, and an account somewhere
unrelated is required for OpenID.

And since I don't want to use any existing account I have (as they are
aleady used for something else), I will have to waste time and resources
and create a new one, preferably at a provider which I'm not using
yet. :/

However I am certainly not willing to add to any individual's burden
just for the, er, pleasure of getting an OpenID for OpenOCD use, so I
thank you very much for your proposal to give me one on you own server,
but I'l go with some mainstream provider.

> > > That can be really difficult if contributions touch on complex issues.
> > 
> > Since the main issue seems to be a lack of "maintainer time",
> 
> Maybe I should have written "issues in the code" to be more clear.

That was clear; and precisely, complex patches require more maintainer
time.

> > maybe a solution could be topic-specific branches (or possibly repos),
> ..
> > not having registered on the OpenOCD Gerrit yet ..
> > I have no idea ..
> > if there are topic branches in Gerrit already that just don't show
> > up in a git clone
> 
> Gerrit does support branches, and in Gerrit you can even think of
> every single pushed commit as a branch.
> 
> The fact that anyone can maintain their own branch in the OpenOCD
> Gerrit is one of its advantages.

From my Motorola experience, I do know Gerrit supports branches :) --
I meant whether gerrit branches were used in the OpenOCD project, and
looking at http://openocd.zylin.com/#/admin/projects/openocd,branches
did not hint that they actually were, nor do they appear in a git
clone.

So either branches can be created in OpenOCD but none exist at the
moment, or branches do exist in OpenOCD and non-gerrit-registered
OpenOCD users like me cannot see them. Which is which?

> //Peter

Amicalement,
-- 
Albert.

------------------------------------------------------------------------------
_______________________________________________
OpenOCD-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/openocd-devel

Reply via email to