Gentoo Linux 2004.0

Posted Mar 5, 2004 16:55 UTC (Fri) by Duncan (guest, #6647) [Link]

Now that you asked... <g> (Beware, in my regular usenet groups I'm known
for my long posts, <g> but also for explanations that allow some folks to
understand the target principles for the first time.)

A practical summary can be seen in the fact that I'm still talking about doing
a Gentoo install, not Zynot, and not running away from both of the as fast as
I can go..

First, some background on my approach and personal views going in..

The research here was strictly for one purpose, that of seeing whether I'd feel
comfortable using and ideally ultimately contributing back to Gentoo (or
Zynot) Linux. I wasn't trying to determine who was right or wrong, or what
someone else's best choice might be, but what my personal one should be,
given that I hold certain views and prefer certain development approaches.

I switched from MSWormOS (a nickname I came up with myself, which itself
says a lot about my views) about two and a half years ago, now, about time I
would have otherwise switched to eXPrivacy (ditto). I'd called myself a
"hobbiest programmer" for some years on the platform, and dumped ten years
on MSWormOS to switch to Linux, because I saw where MS was going, and
the water finally got to hot for THIS frog, who jumped out. This, from
someone who stood in line at midnight for what I now term the Lose98
release, and who ran every public beta for MSIE/OE from version 4 thru 5.5.
(I skipped the IE6 beta as I was researching switching to Linux by that time,
and was spending my time on that so didn't have it to spend on an IE beta,
which by that time I considered a dead end anyway, tho I did install IE6
release when it came out, shortly b4 I switched.)

Since then, a whole new world has opened up. It's at times frustrating how
slow I seem to be on the take-up, given how much there is to learn, and I wish
I'd gotten to it years earlier. OTOH, some general principles such as the
concepts of widgets and windows, and the concept of threading I was recently
explaining when someone asked about threads in response to a kernel article,
have carried over.

My views in the last two and a half years have only gotten stronger. I've been
known to say in regard to proprietary-ware that I didn't dump TEN YEARS
worth of MSWormOS, to come to libre-ware Linux, only to start using
proprietary-ware once again! I'm pragmatic in the sense that I realize for the
time being anyway that the two types of software continue to exist together,
and some may have need of proprietary software temporarily, as they move
toward the libre-ware side longer term. However, make no mistake about it,
I believe in and wish to put my time, money, and energy behind the ultimate
world domination of libre-ware. A few months ago, during the upgrade to
AMD64, I finally removed the last known traces of proprietary-ware from my
computer, including both the Lose98 installation I'd saved, AND the
proprietary-ware NVidia drivers (which would have been removed earlier had
I not needed them to be able to use the second video out on my card, now
replaced with an ATI -- the libreware ATI drivers DO drive both outputs on
a dual-output card).

That covers the views. The environment I prefer for my desktop (server
considerations are obviously different) is one of constant update, as only
possible on open source. I shun the scheduled release idea, save for
"snapshots" which provide a convenient freeze-frame with which to make an
initial installation. Thus, after my initial install of Mandrake 8.1, after some
time familiarizing myself with it, I discovered and immediately began using
Mandrake Cooker, the continously updated alpha distribution.

Now to my research..

The above sounds like Gentoo already, doesn't it? Of course, the thing
Gentoo is famous for is its from-source nature. None of the press about it
says anything about the development philosophy or social contract or any of
that. SuSE's position is fairly well known, as is Red Hat's and Mandrake's,
but until I started my research, I knew nothing about Gentoo's stand on the
libreware ideals I hold so dear at all. This was my biggest initial question.
Would I be comfortable as a user and hopefully eventual contributor to
Gentoo, as someone with strong libreware views? Fortunately, I found the
answer to be a resounding *YES*, at least going from what I saw on the web
site.

However, someone in one of the newsgroups I'm a regular in happened to
mention this fork thing, and knowing that what one sees on a web site isn't
always indicative of the real person or company behind the web site, and
knowing there's always two sides to every story, I began to wonder. It was
clear to me I couldn't continue to increase my interest in Gentoo without
resolving to my own satisfaction what significance this might have to me,
someone interested in ultimately participating in any distribution I choose.

Actually, I didn't have to go anywhere NEAR as far into it as I expected I
might, as I found all the answers I needed directly in the Zynot "Reasons for
Forking a Linux Distribution" document, here:

http://www.zynot.org/info/fork.html

In short form, given my personal ideals and preferred environments, after I
read the entire page, AND the sequal it points to, I found myself asking
"With a presentation like that, who needs enemies?"

I could tell right from the introduction that I'd probably have issues with it,
from the big point he made about being interested from the beginning in
commercially exploiting the embedded side. As I said, I'm pragmatic enough
to realize that proprietary solutions will exist for some time, and I also have
NO problem with someone making money off of Linux and libreware in
general. However, someone putting THAT much emphasis on the money
aspect, when my motivation is on the ideals side, quite apart from any money
I may happen to make doing what I believe in, just isn't going to be someone
I'm interested in following very far. Never-the-less, I both completed that
piece and overviewed the next one, ensuring that I knew his side of the story
and giving him every reasonable chance to change my mind.

Further in, there were several points at which it was painfully clear to me that
his interests and my interests diverged, even while from what else I've read, I
realized his approach might very well fit the embedded market quite well.
(Thus, again, not wrong and right, just "different".)

One, his characterization of all his time being "sweat equity", while Daniel's
view (as Zach portrayed it) was that "many people volunteer their time and
services" outside of any specific contracts. While I can empathize with Zach's
disappointment and hurt under the circumstances, that does NOT change the
fact that I find Daniel's point, even as Zach portray's it, absolutely valid, in
NO SMALL measure because that's EXACTLY the attitude I'd take under
the circumstances, one of "volunteering" my time and energy outside that
which I'd specifically been contracted to work on. Not only do I take that
attitude in regard to Linux and libreware in general, but I EVEN take that
attitude at work, which is unrelated to my HOBBY of computing, and which
I don't specifically care significantly about. I simply believe things work
better in general if everybody works together, and for me, that includes doing
what I can quite outside of being PAID for it, just to keep things running
smoothly for everyone. Yes, I recognize that such an attitude sometimes gets
taken advantage of, but I STILL choose to have it, because I believe in
something bigger, that despite someone's greedy intentions, cannot be taken
away from me. Those that take advantage of such things eventually tend to
disappear due to their own actions or "environmental factors", and if a
workplace and the people in it were sufficiently NOT that way, I'd not be
happy working there and would find someplace else to work. Thus, while I
empathize with Zach, I empathize with Daniel even more, and the hurt he
must have felt finding out that someone he THOUGHT believed in
something enough to put significant volunteer time into it, actually had
ulteriour motives. Of course, for all I know they are BOTH greedy bastards,
but that doesn't really concern me. What does, is that this was a point
AGAINST my participation in Zynot, and either neutral or FOR my
participation in Gentoo, because Zach seems to have no concept of the idea
of "volunteer" work that IMO so contrasts the libraware world from that of
proprietaryware. Many if not most libraware developers do it in large part
because they LOVE doing it, and would CONTINUE to do it, tho perhaps
not to the same degree if they had to support themselves in other ways,
WHETHER OR NOT they got paid for it. The same might hold true to
SOME extent for proprietaryware folks, but to a much larger degree, for
them, it's just "a job", the way the make money, and they do something ELSE
for a hobby. (This, BTW, is why I believe the RIAA and friends are wrong..
Not everybody that sings in the shower has a voice folks will pay for, but the
fact that we DO, without being paid, indicates that creativity WILL survive,
WITH OR WITHOUT the record labels. How many garage bands out there
are there that actually PAY money, for instruments, for gas, etc. in addition
to the time invested in practicing, to be able to perform? Sure, some of these
folks do it only in the hopes of making it big, someday, but you can't tell me
that they'd all stop doing it if the album labels suddenly ceased to exist. I
KNOW, because I compute for the same non-monetary rewards!)

Two, Zach constantly attributes to Daniel and the "gentoo-biz list cabal"
elements of greed and trickery which he himself seems to display. There's the
saying about seeing in others that which reflects from yourself. Now, I can't
judge whether Daniel has that sort of thing planned or not, but what Zach
ironically fails to see is that the GPL itself puts a limit on that -- as
demonstrated by the very fork he's making. If all the allegations ARE true,
then licensing all the Gentoo specific projects under the GPL is a pretty BIG
mistake, because there's no "lock-in". The moment it gets to oppressive, there
can be a fork. Again, the irony of all this is that Zach and Zynot are
demonstrating this very fact, all the while claiming some big conspiracy on
the part of Gentoo and saying how dangerous it is and that they are there to
"save the distribution" as it were. Hogwash! (IMO, of course.) While there
are possibly legitimate questions to be raised about the for-profit
Gentoo-Games side and how it was formed, and Gentoo itself is indeed not
yet a non-profit, the very fact that the GPL was chosen as the base license
says SOMETHING about either their motives or their stupidity <g>, and
one can't argue how Gentoo has grown and the size and at least reasonable
openness of it's organization. The pieces simply don't fit, the way Zach is
attempting to MAKE them fit. Again, even if he's right, the fact that it's all
GPLed means that if it goes far ENOUGH out of the community interest,
there WILL be a fork. It's happenned before, look at the history of SSH, for
example, and THAT was a BSD style license ALLOWING the sponsor to
take it all proprietary. With GPL, that couldn't have happened because all
the community contributed material couldn't have been shifted to proprietary
without the concent of the contributors. As I said, if that's the ultimate nasty
scheme of Daniel and his "cabal", they were sure STUPID when they chose to
put the Gentoo-unique distribution "glue" under the GPL. It just doesn't fit,
and nothing Zach says can MAKE it fit.

Those two items are on the idealistic side. Number three is operating
environment oriented. As I'm sure you know if you've been using Gentoo,
better than I do just beginning to look into it, Gentoo fits the continuing
updates with occasional snapshots model better than the hard "releases"
model that better fits the proprietary-ware world, where development is done
in secret and hopefully more polished "releases" are made periodically for
public consumption and, they hope, for purchase. As I mentioned at the top,
I like the continuously updated model myself. Of course, it isn't so
convenient for the embedded world, where everything must be finely tuned to
work together and assembled into a flash update, therefore naturally fittting
the periodic release model better, so Zach's view here is understandable.
However, I still found it disconcerting to see his emphasis (under "charting a
new course", "business reasons") on..

a) A firm release schedule. A firm release schedule implies a release focused
distribution, something that neither Gentoo nor I seem to be interested in. If
that's his thing, and as I said, I can see how that would make sense for an
embedded focused strategy, my question is why was he with Gentoo in the
first place? I'm beginning to see why the split occurred..

b) A clear roadmap. Linus himself says he attributes part of the growth of
Linux on NOT having such a roadmap, allowing the USERS and
CONTRIBUTORS to shape where Linux goes, based on the needs THEY
have. This seems more like proprietaryware top-down manager-speak than
open source experience. Sure, a roadmap is nice, but open source develpers
aren't afraid to tear it up and go with something else, if that's what the users
are asking for and other developers are submitting patches for. Take a look
at the personal history he outlines at the beginning, tho, and you see where it
came from, he has proprietaryware roots and apparently never HAS gotten
the hang of open source. Thus again, it's not surprising why he eventually
lead a fork.. IMO, he needs to curl up with Eric S Raymond's The Cathedral
and the Bazaar and other essays for a nice long reading session.

c) Prevention of feature-creep. Why? With continual updates and user
oriented feature addition, yes, keeping the "bloat" out is good, but why NOT
introduce that latest feature you have "sort of" working? Remember, release
early, release often, the open source mantra!

d) Follow-through. He says select a release date, and allow only small delays.
Again, why the emphasis on release dates? Where's the continual updates?
Where's release early, release often? Where's the Linux kernel attitude of "it
will be released when it's ready" in regard to REAL milestones? (Such
milestones IMO aren't really the domain of distributions anyway, except for
such things as adding support for additional architectures, but rather the
domain of the projects a distribution assembles, KDE, X, the Linux kernel,
etc. thus the continual updates model for a distribution as the individual
components within it update or release THEIR milestones.)

Obviously, by this point, it's becoming quite clear that Zach and by extension
Zynot and I could NOT be MORE incompatible. One has to wonder why
he's in the open source community AT ALL, and ESPECIALLY Gentoo,
given some of his beliefs and attitudes. No WONDER he's forking! It's also
becoming clear that where he disagrees with Gentoo, I disagree with him,
making me even MORE confident about my initial Gentoo optimism.

That said, many of the points he brought up about communication and
delegation and fuzziness of domain apparently DID need to be addressed,
and his actions WERE instrumental in bringing reform. From my point of
view just coming it it's hard to say, but reforms WERE made, altho, naturally
some changes would HAVE to be made with someone that big pulling out of
the project.

As well.. in the second document there's a FAQ of sorts where the question
is asked what Zynot's reaction to a fork might be, under the same
circumstances. Zach says something there that I believe makes sense, and
that I'd like to see in more projects, much as it's done in the kernel. He says
he envisions and would like to encourage various "mini-forks" and gives some
examples of where it might happen. He says given this "mini-forking", while
still keeping everything within the same cooperating family, it'd be his hope
that the need for a full major fork is averted. I'm not sure Zach understands
the significance of that idea, given how he evidently misses some of the other
big differences between open source and proprietaryware, but..

As I said, this is exactly the model the Linux kernel takes. Eric Raymond in
some of his essays points this out as well. While MOST open source projects
and the culture give one the RIGHT to fork, in practice and as the culture
actually works, ACTUAL forking is generally HIGHLY discouraged, and
taken as a sign of disrespect and offense re the current project leader. This,
Eric says, could be one hold-over from the proprietary and strong core
developer team era, before the full implications of the advantages of open
source were fully understood.

Even within the stalwart open source community as lead by Stallman and
others, there wasn't a recognition of the TRUE power of open source
including the many. Another example here is GCC, which was while open
source, developed by a small core group for many years. Along came a group
wanting to open up development, and a somewhat friendly fork occurred,
with the original GCC still being developed by the original core, and a new
EGCS taking wider community contributions. Eventually it became painfully
clear how much faster the community EGCS version was developing than the
traditionally small core developed GCC, and the projects merged again, with
EGCS becoming the new GCC 3.x, and the new more open community
development model continuing. (Note that I really got into Linux and
libreware only shortly after this, so I can't vouch for the accuracy of the
above, but am going on my understanding of what I've read, including Eric's
essays.)

Back to the Linux kernel. Part of the inspiration for EGCS was based on
observation of the Linux kernel and how amazingly fast IT evolved, even by
open source standards. As mentioned above, Eric attributes part of this to
the "mini-forking" not only allowed, but actually ENCOURAGED, with the
Linux kernel. Linus or his hand picked version lieutenants put out the only
"official" versions, yes, but there are all sorts of -ac and -mm and etc
mini-forks, besides the versions of the various distributions, and the testing
space is made effectively that much broader. Each "fork" has its own publicly
available patches which as they are tested gain adoption in other kernels,
getting even MORE testing, until the best of them get added back into the
official kernels. This is ON TOP of the already-faster-than proprietary
evolution of the main kernel and its normal testing and feedback loop,
compounding the evolution rate several times over.

This is just my opinion (and that of others like Raymond), but it seems a
shame to me that so few open source projects take advantage of the same
compounded evolution rate with the same techniques. It seems to me (and
others) that friendly forking a la Linux kernel should be ENCOURAGED,
not DISCOURAGED, and that the strong open source cultural repellance to
the idea of forking, even under friendly terms, needs to be reexamined. Thus,
seeing Zach's mention of "mini-forking" DID strike a chord in my mind, as its
something I'd like to see more of. There, I agree with him. Unfortunately,
he mentioned nothing about the kernel comparison or what happened with
GCC/EGCS, nor did he mention ESR's essay on the subject, so I must
believe he wasn't familiar with the idea nor had he really explored it or
anything ELSE, it would seem, in open source, in any significant detail. Still,
I gotta call it when I agree with his idea.

.. Like I said, I have a reputation for long posts. Hopefully, my reputation for
clarity of explanation also comes thru in this one. <g> (And yes, it DID
take me some serious time to write that... <g> However, as is generally the
case, writing it helped clarify things in my mind as well, so it wasn't a waste
even if nobody else appreciates it.)

Duncan