directory-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Colm O hEigeartaigh <cohei...@apache.org>
Subject Re: Apache Directory and SCIM
Date Wed, 22 Apr 2015 10:42:52 GMT
I don't think that having two different implementations of the same thing
at Apache Directory is a good thing. However, it's certainly not unheard of
for a project to replace an older codebase with a new donated codebase for
a new major release, if there is a compelling reason to do so. For example,
maybe the architecture of the older project may be too limiting etc.

Is this an option? Is your project demonstrably superior to the existing
escimo code?

Colm.

On Wed, Apr 22, 2015 at 11:28 AM, Steve Moyer <smoyer@psu.edu> wrote:

> I think we've spent enough time rehashing the past ... let me just say
> that I think things would have gone a lot differently if there'd been an
> easy way to provide a repository branch for us to check code into.  On
> the other hand, I can certainly understand why you don't want to change
> your infrastructure every time someone pops into the IRC channel - SVN
> doesn't make it so easy to limit permissions and you probably see a
> hundred people a month come by then disappear (like I seemed to do).
>
> I'm also not debating the use of the work forked ... just pointing out
> that Kiran and I decided *together* our approaches, while both valid
> ways to tackle the problem, were not compatible enough to live in one
> project.  There was a reason we went looking for an OSS Java-based SCIM
> project ... we can chat on IRC in a couple hours (once I'm in the office).
>
> Assuming Shawn (my boss) still wants to push forward with Apache
> Directory, there is the issue of having two SCIM implementations in one
> project ... I guess the value propositions of each would need to be
> clearly stated?  I'd like to see wide-spread SCIM adoption, so we
> certainly don't want to spread any confusion.
>
> Steve
>
> On 04/22/2015 05:03 AM, Emmanuel Lécharny wrote:
> > Le 22/04/15 03:55, Steve Moyer a écrit :
> >> Kiran,
> >>
> >> I certainly didn't want or expect you to be disappointed!
> > Well, as human beings, we have feelings, which may be overstated at some
> > point. DIsapointement might be a bit strong : it was more a surprise to
> > me than a disagrement, as I really thought that the discussion we had in
> > 2013 (JavaOne) was quite constructive, and it somehow died soon after.
> > You might not be the only one to 'blame' for that !
> >
> >>  It is however
> >> a bit disingenuous to say that *WE* forked eSCIMo and kept it private
> >> since you and I discussed the fact that your ideas for eSCIMo moved
> >> towards an untyped mapping system while we moved towards a fully-typed
> >> model of the SCIM resources based on a first-principles dissection of
> >> the SCIM specification.
> > Again, when I wrote 'forked' I mean it was not part of the repository we
> > had back then. Technically, 'fork' is the exact term, there is no
> > positive or negative implication here : you simply started from some of
> > the existing eSCIMo code, and deviate from it. There is nothing bad
> > here. This is Open Source for this exact reason.
> >
> >> It was your suggestion that we fork the project (at the time I think
> >> there were three files checked into eSCIMo) and the major contribution
> >> to the project is that it's still (for the time being) using Wink.  You
> >> also suggested the name "igloo" and didn't feel the Apache Directory
> >> server project would want to create another repository for it.  Since we
> >> had a Gitolite system running (and hadn't created the PennState Github
> >> account yet, we added your public key to our Gitolite system on July
> >> 15th, 2013 and gave you read/write access to both the igloo source code
> >> and the kerberos-client source code (which is not nearly finished).
> > My memory is a bit vague. You must be true. More later...
> >
> >> We've only had sporadic conversations on IRC and I personally could have
> >> done a lot more to keep this team up-to-date.
> > That's probably the problem here. It's all about perception, and as we
> > are all pretty busy on other matters, being disconnected create the
> > false feeling that every group is working on his own side of the
> > internet. All that I recall is a few exchanges (technical ones) in
> > novemeber/december 2013, and nothing on IRC (I keep all my mails and IRC
> > convos, just in case ;-).
> >
> > What would have been way better is to have welcomed you from the very
> > beginning, and use the Directory repository as a base for yoru code.
> > That would have let other committers (and specifically Kiran) to be part
> > of the effort.
> >
> >
> >>  Igloo is *NOT* back-end
> >> specific but our implementation, which is plugged into igloo as a
> >> provider is very specific to OpenLDAP and Fortress.  It's also not
> >> completely SCIM compliant.  I guess the biggest reason I haven't spent
> >> more time collaborating on igloo is that a) you didn't seem interested
> >> and I knew that eSCIMo was developing as part of a delivery to your
> >> client
> > There are no client for eSCIMo. We don't even have a release yet. Kiran
> > most certainly started to work on eSCIMo as an effort to implement SCIM
> > because one of his client needed it, but in any case, the effort was
> > certainly not client oriented. What happens, generally speaking, is that
> > once you have started something that fits one client, it's likely to fit
> > another one, or many other ones.
> >
> > Regarding my interest, I don't remember having given this impression :
> > as eSCIMo has been started as a subproject, I would have *loved* to see
> > more people to be part of the effort. You were interested, and you would
> > most certainly have been a great addition to the team, and AFAIR, I told
> > you so (but again, two years later, it's easy to rewite history ;-)
> >
> >
> >>  and b) we had our own dead-lines for completing the implementation.
> > That I can understand.
> >> In the mean-time, I talked with Emmanuel at JavaOne 2013 and Shawn
> >> McKinney at JavaOne 2014.  Emmanuel never sent me a public key for the
> >> repository but even today,
> > Not sure you even asked me one.
> >
> >>  you can check the igloo source out of our
> >> repository.
> > This is not the way it goes. We don't have time to check what is done
> > outside of Apache, and we aren't fishing for side projects to join us.
> > It's philosophically not the way we proceed, for many reasons. The main
> > reason is that we want committed people, who will sustain the project in
> > the long run. That also means the existing community is ready to accept
> > the project - ie if the original contributors leave, the community will
> > still be around to keep it alive -. So it's for everyone outside Apache
> > to express a will to join us, not the other way out. We can tell people
> > that what they are working on would be a great addition to The ASF, but
> > we won't go any further. More critical : in order to feel comfy working
> > in such an inclusive community, one has to understand what it implies,
> > and it implies a lot.
> >
> > So to speak, the very day you seems ready to jump into the band wagon,
> > assuming you accept the pros and cons (and there are many cons !), it is
> > a joy for us to welcome you.
> >> In any case, please accept my humble apologies ...
> > No apologies needed. You haven't do any harm. We just have a
> > miscommunication issue, nothing more. It happens, it's not your fault,
> > it's mainly a mismatch that can be solved.
> >
> >>  I don't want the
> >> chasm that appears to have grown between us to grow any wider!  Can we
> >> at least build a bridge?
> > We don't need a bridge ! There is no river, we are not that disconnected
> !
> >
> >> I won't be at all insulted if you reject the idea of adding igloo (or
> >> whatever it ends up being called) to the Apache Directory project.
> > Certainly not my intention.
> >
> >
> >> SCIM
> >> has a relatively large following and once the code we're offering is
> >> cleaned up, documented and fixed to match the final SCIM 2.0
> >> specification (we're currently bringing it up to draft 17's state), I'm
> >> sure the code will be appreciated by the SCIM community and could stay
> >> either on the PennState GitHub repository or be incorporated into the
> >> Internet2's middle-ware offerings.  Since the individuals reading this
> >> e-mail have been hugely helpful in our meeting the university's goals,
> >> we thought it was only fair to offer it back here where it started.
> > That's certainly a good thing.
> >
> >> This is also not a code dump - We expect to spend a significant amount
> >> of time bringing the code up to standards (our own and if necessary,
> >> your style guidelines) and we're actually going to fix the architectural
> >> mistakes made in the code before the world starts using it.
> > I guess that Kiran used 'code dump' for some reason : we have had to
> > deal with a load of such when we reworked the Kerberos code, which was
> > literally a code dump. And we did that the year before we met in SF.
> > having spent 6 months literaly rewriting the full Kerberos code was not
> > my best experience, not was it Kiran one...
> >
> > I would not have called your proposal a code dump.
> >
> >> I expect to
> >> remain a committer and perhaps the primary maintainer whereever the code
> >> ends up (even if we keep it private).  Fortunately, we've been hiring
> >> some great JavaEE Software Engineers in the last few months and I'll get
> >> to go back to being the architect/programmer.
> > Good !
> >> We'll also be releasing a pretty broad suite of tools along with the
> >> framework - this e-mail is long enough without listing out the modules
> >> we'll provide.
> >>
> >> In any case, I'm truly apologize for any hurt feeling this has caused.
> > Bah. Don't be sorry, I was just trying to express some of my feelings,
> > and I probably overeacted too. For a miscommunocation, you need at least
> > two persons, and I'm the broken piece on this side of the communication
> ;-)
> >
> > Let's reset this full convo and restart from your original mail from now
> > on !
> >
> > Emmanuel
> >
>
> --
> "One longs, in reading your code, to walk on all fours" - Voltaire
>
>


-- 
Colm O hEigeartaigh

Talend Community Coder
http://coders.talend.com

Mime
View raw message