[oi-dev] oi_151a9 roadmap & planning

[Garrett D'Amore]

On Sep 3, 2013, at 2:11 AM, Joerg Schilling <Joerg.Schilling at fokus.fraunhofer.de> wrote:

> 
> I am not sure whether you are aware about the fact that OpenSoaris has few 
> contributors and that Illumos is trying to frighten developers by not being a 
> sure partner you may trust. 

Aargh.  I hate responding to Joerg posts, and I know I probably shouldn't, but its not in me to ignore someone who goes around throwing rocks at my house…  and there may be folks who are not yet familiar with Joerg's antics, so I feel that some response is unfortunately necessary.

> 
> Promising to include software causes people to spend time in creating a webrev, 
> but later not doing a code review causes at least this time to be wasted. This 
> finally ends up in not fulfilling a commitment and is toxic for Illumos 
> credibility.

This all comes back to star(1).  And you're continuing with your antics that really result from the fact that you're angry because I refused to personally review your software after I had such negative interaction with you.  I told you then, and I repeat now, all you needed to do was find another sponsor willing to review your work, and you *were* required to have your code properly reviewed if you wanted it to be integrated in illumos.  You've never undertaken that, and your refusal to follow the rules that we have established for *all* contributions is why your software isn't integrated.  Further, going around doing what you can to disrupt illumos operations has turned off almost anyone else who might be wiling to work with you.  You urinate on everyone who might have been wiling to give you some help, and wonder why nobody wants to?

> 
> Adding code with known bugs to Illumos even though it did not pass a 
> codereview, just because the code is from an animal that is more equal 
> than others destroys credibility.

No.  This has not happened.  I believe you are referring to a hexdump / od incident, again, where your sum total of code review was "its broken" with nothing actionable, except that you were again upset because we were not going to integrate hexdump which you had hastily put together an od-workalike mode for (instead of simply providing actionable feedback to my already complete program.)  Other reviewers passed the review and to this date I know of no actual defects in od(1) -- feel free to file bugs if you find otherwise.  (I also *did* give you review feedback on your version of od(1), if you recall, and that was actionable.)


> 
> Illumos also drives in a direction that is not what I understand by going the 
> Solaris (UNIX) way. Code is removed just because it is not of interest for 
> Nexenta or Joyent, continuing this way will at some time lower attractivity.
> Ripping off troff from man(1) because "col -x" (as a result from a buggy 
> localization software in Illumos) does not pass japanese characters is a 
> strange response to a problem. 
> 

man continues to support roff functions, but we also support mandoc now, which is a much more elegant formatting packaging.  It also makes it vastly easier for us to collaborate with *BSD on man page content.  The fact that col -x was used for formatting was just one more level of brokenness in the existing package, and we made a decision to switch directions.  There is no collaboration from the *Solaris* ecosystem anymore, and so we don't consider that important.

> 
> 
> While points 1..4 are decisions that can be made by OI, your proposal is not 
> leading to a decision that could be made by OI.
> 
> You may hope that this at some time is handled by Illumos, but Illumos does not 
> cover all the skills that are needed. So my question to other OI members is: 
> are you 100% bound to Illumos?
> 
> I am currently underway with finishing the package collection for SchilliX 
> based on network loadable Svr4 packages that I made basically working in 
> February 2011. Now that SchilliX collects helpers, we expect to be in a state 
> to deliver a full desktop in a few weeks. After that, I will start to work 
> again on the OpenSolaris code base. As mentioned, a toxic person prevents 
> Illumos from being interested in collaboration, so we either end up in a 
> tattered OpenSolaris landscape or we find a way to pool changesets and let 
> distros decide how to compose their OpenSolaris base.
> 
> Are you interested in collaboration?
> If yes, do you have an idea how this could be done?

I strongly caution the OI team to consider carefully… Joerg has a very different idea of collaboration than the illumos team.  Once you go down the road of SchilliX, you'll be departing further afield from the OS that the main contributors to the kernel (Joyent, Nexenta, Delphix, DEY, OmniT all employ people to do this for a living) use and maintain.  And unlike illumos, where the rules for integration are clearly spelled out, and the leadership is indeed distributed, you'll be entering what is a one- or two-man show, where Joerg's word is law, and the other contributors are folks who couldn't operate within the framework for contribution that illumos has set.  (E.g. Martin Bochnig.)

	- Garrett