[nSLUG] Jigdo File
synrg at sanctuary.nslug.ns.ca
Wed Mar 5 14:31:03 AST 2003
On Wed, Mar 05, 2003 at 01:30:37PM -0400, Donald Teed wrote:
> Ben Armstrong wrote:
> > Readjust your set. It does not mean broken. This is a misconception
> > perpetuated by dozens of whipped-together projects that appear on freshmeat
> > and elsewhere that really never should have been made public in the first
> > place.
> No, < 1.0 = beta it is a conception formed from using Beta versions
> of many packages, including Netscape circa 1994, Internet Explorer
> when it was still known as Spyglass, NCSA Mosaic, and more recently
> with open source tools.
Where does what you say here contradict what I said? In your case, the
conception may have been formed by those particular experiences, but in
general, I believe it is also perpetuated by freshmeat and other sites
offering or packaging new software (including, shock, horror, Debian).
There is an alarming trend in Debian that if a package exists and it looks
cool, we must package it. This gives users a bewildering choice of broken
pieces of software to install. With projects like Debian Jr. I'm fighting
that trend, trying to zero in on a subset of everything Debian offers that
is useful (and solid) for a particular user base.
> I don't spend any time at freshmeat - my open source beta
> experiences are almost purely from packages provided with stable
> releases of Linux.
Stable releases of Linux include a lot of stuff some developer packaged
because it appeared on freshmeat and they thought it would be cool to
package it. (Speaking of Debian here -- I cannot speak for other distros,
although I suspect this is true in varying degrees for many of them.)
> I don't know where this "private meaning" is coming from. What I'm
> talking about with version numbers is common practise in all
> commercial software vendor releases. I can't think of one
> for-profit company that sells software at less than 1.0 (well,
> at least they would chase any early sales with updates/patches).
"Private meaning" in that you assert what 1.0 means, and then assume that
everyone else takes it to mean that too. There's a good reason companies
don't sell software less than 1.0, and that is marketing. They want the
appearance of having produced a "finished product" whether that is true or
not. It is all smoke and mirrors. A volunteer-run organization like Debian
does not have this marketing pressure to make all version #s of material
they release in stable releases >= 1.0. And I stand by that. If the
material meets certain quality standards for release (i.e. it at least has
no release-critical bugs filed against it when release-time finally arrives)
then it should remain in Debian. If not, it should be turfed. Nobody at
Debian agonizes over what the version#s are and hounds upstream to change
their version#s to meet marketing requirements. I would be shocked and
dismayed if they ever started doing that.
> There are some packages that seem to never reach 1.0, and in those
> cases I'd say the developer may have goals that are beyond reach, or
> they just don't want to take on the responsibility.
And that's just fine. That's why I exhorted you to readjust your set.
There's no point in railing out against the free software development
community about taking too long to go one-point-oh. It's part of the
culture, and it is unlikely to change, for the reasons I have already
> No one has a specialty in releasing buggy software. It is everywhere,
> no matter what the political stripes. The reasons why it is
> released in that state are different, but the buggy packages
> are everywhere - open source and commercial.
So far, you are arguing *for* my position here. If buggy software is
everywhere, open source and commercial, and if companies *never* sell
anything at less than 1.0, and < 1.0 means "not fit for general consumption
yet" then there is a whole lot of whitewashing going on, because there are
plenty of commercial packages out there that fit this description.
> There is little point to this type of "honesty" when we are forced
> to use a package in early adoption as with the case of jigdo or
> in the case of mod_perl 1.99 and Apache 2.0 provided in Redhat 8.0.
> I'd rather have the option to download the full ISO than be forced
> to find Woody via the busted jigdo.
Woah, back up here. Jigdo remains the best option for a certain subset of
Debian users. From those who I know personally, I have heard very few
complaints. The fact that you've grappled with bugs in it is unfortunate,
and it may well reflect the uncompleteness of the software. But who made
you any guarantees that it would do what you expected? And who forced you
to use Jigdo? Over the history of my use of Debian I have used a whole host
of different ways of installing Debian, and the most tried-and-true for me
- install from single CD produced by a company like "Cheapbytes" (i'm not
current with this; last time i did this was a few years ago -- Don't even
know if Cheapbytes still exists)
- small base-install floppy set + network install for the rest
- even smaller rescue/root pair of floppies + network install for the rest
Sure, that doesn't cover all possible users' needs, but they *are* options,
and good ones in my experience.
> > some consistency?" Bah. Who is "we"? When I volunteer to develop free
> "We" is the entire community of open source software users. If the
> users don't exist, the developers might as well be playing Starcraft.
So you are now a spokesperson for the entire community of open source
software users? I'm sure you have some good ideas about what users need,
but I am hoping "more consistent version#s so we can tell what is good and
what is not" doesn't rank very high up there in general. It's just not
terribly realistic. I'm sure we'd all like version#s to mean more than they
do, but as is evidenced by the boatload of crappy software out there >1.0
and good software <1.0, that just ain't gonna happen. You'll get far more
reliable indicators of what's good and what's crap in watching bug reports,
vitality indices, and most importantly the comments of other users in user
support mailing lists then you *ever* will by watching version#s.
> "We" depend on reliable software to get through our day and not come out
> looking like fools to our bosses who gave us the option to use whatever
> toolset we thought was most productive. When I fought with jigdo
> for a couple of hours, that wasn't productive. Perhaps this is
> something that needs to be remembered. Linux isn't just about
> your home computer giving the boot to Windows. It is used
> in business, and in that context the priorities can be different
> and assert a different set of values.
At some point you should have (and may have, for all I know) considered
jigdo a flop for your case, and just moved on to another option. There are
several, as I have mentioned. Or with perserverance, you might have been
able to file some coherent bug reports and have seen some action on them,
benefitting everyone in the jigdo user community. That's entirely at your
discretion, depending on the amount of time, energy and inclination you have
for that sort of thing. Or if it really is as crappy as you say it is, you
should've been able to easily find the litany of bugs filed against it or
complaints in the mailing list(s) that support it to warn you away from it
in the first place, saving you the embarrassment that you suffered in front
of your boss.
> I don't want to be forced into using something with poor
> reliablity, and that is what happened in the case of jigdo.
I don't buy that you were forced.
> Perhaps the version numbers mean nothing in betaness in the open source
That's what I have been telling you.
> Perhaps that is how something as unreliable as jigdo
> became a mandatory application for Debian ISO retreival.
It isn't. It's a work-in-progress to try to address the problem of
selecting a user-tailored subset of the large complete set of ISOs that is
the default offering. Two viable CD-install alternatives are:
1. use the full set of CDs
2. find a reliable single-CD or two-CD set someone else has already
constructed and use it instead (they aren't that hard to find)
> Perhaps the open source developer community (including commercial
> vendors like Redhat) need to rethink how beta software is deployed.
What, go back to the "cathedral" model of software development where the
development team spends a huge amount of time behind closed doors hammering
out the software, only to release it late, over budget, and full of bugs
anyway in the end? No thanks. Or perhaps you had something else in mind?
By designating software "beta" and releasing it I am saying "I'm not done
with this yet and I need your feedback, so please download it, try it, and
if you have any problems, file bugs." A lot of software, as you complain, is
in this state. Not all of it is < 1.0, either. Software is constantly
going through alpha and beta cycles for each release, regardless of the
version#. In other words "betaness" is totally independent of version#.
We all wish that software were released bug-free. We've all experienced the
pain of struggling with an application we had high hopes for and were let
down. But you have mis-identified the problem. It isn't "it's too bad
we're forced to use all of this < 1.0 version software". It's more like
"it's too bad it's not easier to find what the best option is when I'm
looking for a package that will do 'foo' for me." That remains the key
problem for Debian users and for users of software everywhere, commercial
and non-commercial, non-free and free. The answers for now are: "do
thorough research before you even download," and "when you are let down by a
piece of software, know where your resources are so you don't have to fight
against it alone." And that's about all I can say that could possibly be of
help. I really am sorry jigdo was such a bad experience for you. Now I
hope I have illuminated some of the real issues for you.
Disclaimer: I am not, and probably never will be a jigdo user. I have no
specific expertise as a jigdo developer or user. I argue merely as a
developer and happy user of plenty of pre-one-point-oh software that I am
very proud of and satisfied with.
,-. nSLUG http://www.nslug.ns.ca synrg at sanctuary.nslug.ns.ca
\`' Debian http://www.debian.org synrg at debian.org
` [ gpg 395C F3A4 35D3 D247 1387 2D9E 5A94 F3CA 0B27 13C8 ]
[ pgp 7F DA 09 4B BA 2C 0D E0 1B B1 31 ED C6 A9 39 4F ]
More information about the nSLUG