Re: Packaging and installation
On Wed, 25 Oct 2000, Nicholas Petreley wrote:
> But this standard is most definitely called for. Who here hasn't EVER
> been frustrated by a long string of unmet dependencies when they knew
> darned well that some of them were met on their system? Who hasn't
> been frustrated by the fact that a program complains that it can't
> find libwhatever.4.2.0, and didn't know where to get that library?
> Honestly. Anyone? Did it matter to you at the time whether or not
> that library might fall into the narrow list of libraries covered by
Nick, I want to point out that I believe your point of view is skewed.
You're apparently viewing this debate entirely from the point of view of a
GNU/Linux _user_. The LSB was not really formed to address user concerns.
Distributions are generally concerned about user experience. The LSB was
created because distributions and ISVs wanted a way to have a standard to
facilitate portability of software across distributions. To even the
field so that one distribution cannot dominate and to facilitate software
development on Linux.
Now, a distribution that is LSB compliant and hasn't been screwed up by
the user WILL be able to install LSB-compliant packages. That's the whole
point of it. If a distro is LSB-compliant, it WILL have the libraries
needed. There will be software packages that will require libraries not
covered by the LSB. Now, I'm sure there a going to be a lot of programs
that won't be LSB-compliant. Most of these can address their problems by
statically linking the libraries they need or including them. Is it
perfect? No. But it's a necessary first step towards future specs that
can be more inclusive.
Now the other thing that bugs me about your arguments is that now you're
arguing about something different. Now you're wanting to discuss the
selection of libraries covered by the LSB (which is yet _another_
What are we talking about here? Package management API? Libraries?
Scope of responsibility of the LSB? Purpose of the LSB? This discussion
is unproductive, since we cannot seem to actually discuss issues point by
I suggest that if you want to continue this discussion, that instead of
posting disorganized comments about the lack of libraries, your
dissatisfaction with the package format, and your feeling that the LSB
should grow beyond its charter, that you post a single document on each
issue, describing concisely and concretely what you perceive as a problem,
and give a proposed solution. And by solution I don't mean something like
"use an API". Please don't take this as me dismissing you -- I'd love to
consider your ideas and get a decision made, but it's not going to happen
without a documented proposal.
I noticed that you are going to work with Anthony on fleshing out his
ideas on Chapter 13, and that would be great. We can work with that.
But your comments on the scope and purpose of the LSB, and now your
concerns with the lack of libraries in the standard -- we can't work with
that without documentation of your ideas. If you feel more libraries
should be added, then tell us which ones. Most likely you will mention
libraries that had been considered before, but rejected. Any document you
submit will need to address the conclusions reached in the past.
I know this is a lot of work on your part, and well, it SHOULD be. I
always think of the old saying "too many chiefs, not enough indians" when
I read discussions on many Free Software lists. Posting ideas is
wonderful, but if they are in opposition to the decisions made by the
folks doing the work, you need to put forth the effort to explain your
position thoroughly and/or implement it yourself.
I've heard some great ideas from you. I like all of them but where I
disagree with you mostly is in the practicality of doing them.
I really look forward to reading and debating something other than ideas.
| Jeffrey Watts |
| firstname.lastname@example.org o-----------------------------------------o
| Systems Programmer | "The strength of the Constitution lies |
| Network Systems Management | entirely in the determination of each |
| Sprint Communications | citizen to defend it. Only if every |
o----------------------------| single citizen feels duty bound to do |
| his share in this defense are the |
| constitutional rights secure." |
| -- Albert Einstein |