Dan Price wrote:
[snip]
> That aside, I don't plan to engage in your-patch-vs-my-patch.  At a
> practical level, our processes are not sufficient to mitigate a dispute
> between two competing implementations of the same technology.  More
> importantly, I support some of the high level goals which Roland's
> project is already pursuing, such as the goal of having ksh93 in the
> OpenSolaris project.

Are you talking about "OpenSolaris" or "OS/Net" in this case ? =:-) Note
that I started this project with the clear goal of integration into
OpenSolaris while thinking (and many others thought the same way or
still are not aware that there is a difference) about "OS/Net".. I
really wouldn't have started the project at all if the only goal would
be the "inclusion into SFW". One of the goals was always integration to
"OS/Net" and then - step by step - start to replace the old ksh with
ksh93 at consumer level and finally migrate /usr/bin/ksh to ksh93 and
maybe even /sbin/sh (as a follow-up project).

And to make it clear (I wrote this elsewhere, too): We will not make the
shared library interfaces public before ksh93s (="ksh", spec="93",
version "s") or later, but use it in OS/Net long before that point. IMO
that alone justifies an integration into OS/Net - otherwise OS/Net would
depend on non-public interfaces which live in a completely different
consolidation (and likely under completely different and less strict
review/mainataince/etc. scheme). 

The problem right now is that we "only" add ksh93 to OS/Net without any
other dependicies to avoid complexity in the ARC case or causing pain
for anyone else (this is why I am very opposed to a "flag day" - if we
go and hurt other people with the libcmd thing we could enable all the
other more complex changes like the kernel module to support compiled
shell code, too. But I am not really happy with that, I usually prefer
to avoid turning other people's day into a living hell when there is no
compelling reason for that...) which generates lots of attack points for
the "ksh93 in SFW" fraction - and any argumentation against it needs to
be done with care because the wrong argumentation generates even more
debates and questions.

I think at some point the SFW discussion needs to stop - I answered more
emails about the "ksh93 in SFW" debate in the last two weeks than I
wrote or answered any ksh93-related emails in the last three months (!!)
and if this kind of load continues I am simply going to croak (note that
I am a volunteer and I am not even working for Sun which means I have to
work for food elsewhere and my resources (including "free time") are a
rare resource). I really really like to return to hacking code
somehow...

> So I have no interest in trying to take the
> project away from him.  And finally, I appreciate his efforts to
> trailblaze and show how projects run by community members who don't work
> for Sun can succeed (and certainly he will provide valuable feedback
> about what was broken about the process and what needs to be
> streamlined).

Well, that's already on my ToDo list to summarise some of the issues
we've hit. This starts with simple items that we need a BugZilla (and I
am asking specifically for "BugZilla" from bugzilla.org in this case) to
organise things (to avoid large amounts of private email traffic which
should have been done in a public bugzilla), the problem that Sun still
tries to run an OpenSource project like OpenSolaris.org with what I call
"company rules", that integration should be done much faster and in
smaller pieces (the last item is actually very important since I am not
aware of many volunteers who are actually able to wait that long for a
putback, almost everyone else I know would have left the project in
anger (which is an understatement) long ago) and a couple of
Sun-speficic issues which shouldn't be discussed in the public right
now.

----

Bye,
Roland

-- 
  __ .  . __
 (o.\ \/ /.o) roland.mainz at nrubsig.org
  \__\/\/__/  MPEG specialist, C&&JAVA&&Sun&&Unix programmer
  /O /==\ O\  TEL +49 641 7950090
 (;O/ \/ \O;)

Reply via email to