[oi-dev] Resignation as OI Lead

Alasdair Lumsden alasdairrr at gmail.com
Sat Sep 1 00:31:31 UTC 2012


Hi Milan,

On 30/08/2012 08:46, Milan Jurik wrote:
> Hi,
>
> it is not good you resign. And it is not good to see OI going nowhere.
> Here I had long e-mail about how stupid is to think OI is threat to
> Nexenta (Oracle way of thinking), how irrelevant is what Bryan Cantrill
> says, how stupid is to give up developer desktops etc. But nobody would
> read it and it is not important now.

I'm sorry for resigning, but I was sacrificing considerable amounts of 
personal time and energy to the project. It couldn't continue indefinitely.

I was hoping that we could set up a framework for others to contribute 
and carry things forward, but there was just too much baggage to deal 
with. OmniOS's approach of starting from scratch was the easy way out, 
something we couldn't do.

I'd say illumos-userland was the final straw. It took the wind out my 
sails, especially with the "working set" farce where Bayard absorbed 
everyone's changesets then disappeared. The whole illumos-userland thing 
kind of pissed everyone off

> More important is - what is immediate impact of your resignation? Is it
> end of OI?

It is not the end of OI, as long as others want OI to continue and can, 
for example, help Jon Tibble with his efforts, something I may do if I 
can find time.

But I was unable to muster the time/energy to lead the project, so 
continuing as lead was leading the project no-where. I had to resign to 
make way for someone else - I hope someone will step forward, or that 
the project will continue without a leader.

>  From my point currently Jon is moving /dev slowly forward - is he
> willing to continue in it with help from few remaining people?

I hope so.

> oi-experimental is dead end, so is illumos-userland. Good thing. It was
> big switch breaking things and wasting resources.

Well, it wasn't that big a switch - Solaris 11 has userland-gate, and it 
was just that. Andrzej Szeszo has the latest JDS built along with the 
latest userland-gate running on his desktop PC at home. He managed to 
get it all built and working. It was unfortunate he was unable/willing 
to upstream his work. It was possible, though, Andrzej's work proves 
this. We just needed talented, committed developers with time and energy.

> oi-build is slowly taking attention as much better way in our current
> situation.
>
> There is question about prestable, if it should stay as prestable or we
> can accept it as moving dev (as it is now in reality). Would automated
> release every 2 weeks help? Yes, some releases can be broken but we have
> BEs and people will be forced to fix things or leave.

I don't know. Jon's reservation about releasing his prestable builds as 
stable was that there was still loads of CVEs in it.

Automated releases would improve things but there would be a lot to 
automate.

I feel oi-build as a framework is the future, but we tried to jump to it 
too quickly. An incremental approach based around Jon Tibble's work, 
where packages are moved one by one from SFE/JDS/etc into oi-build would 
fix that.

There are however some big challenges. We no longer have the in-house 
skills to rebuild JDS. With your work on SFE and expertise with 
spec-files, perhaps that is something you could help with?

> Currently I fixed the most of things around OI SFE so I can move my
> attention back to illumos and OI after my vacation the next week. With
> small change, from July I switched my position in company and my spare
> time is very limited. But long winter nights are comming :-)

I think OI would greatly benefit from your input Milan and I'm pleased 
you're considering helping it.

> Best regards
>
> from mad man who is using OI as his primary desktop at home

You're not alone :-)

Alasdair




More information about the oi-dev mailing list