[mythtv-users] Poll: apt-get FC3 install

John Van Ostrand john at netdirect.ca
Thu Dec 30 19:49:41 UTC 2004


On Thu, 2004-12-30 at 17:51 +0000, Martin Ebourne wrote:
> On Thu, 2004-12-30 at 10:41 -0500, John Van Ostrand wrote:
> > That is a problem between keyboard and chair but not at my desk. It's
> > the package maintainer.
> 
> Not in this case I suspect.

I took a fresh, stock installation of FC3, followed Jarod's MythTV
HowTo, ran the apt-get mythsuite and encountered dependency issues using
apt-get then who's error is it then? I'm certain I followed the steps,
there were quite easy.

> >From your previous email:
> > I tend to bend the rules from time-to-time, installing
> > something from a tarball instead of an RPM so I use rpm's --nodep
> > after which apt-get seems to refuse to work.
> 
> The answer would appear to be staring you in the face, but you just
> don't seem to see it.

See above, do you still think that way?

> > And yes, my main complaint about apt-get is that it can not keep up with
> > oddities. I could not find an option to override certain checks, or to
> > allow for event the smallest inconsistency. Rpm has it (--nodeps, --
> > force, etc) then shouldn't the tool that sits above rpm support
> > something similar?
> 
> If you aren't able to fix the mess you create when you use --nodeps or
> --force, then don't go there. Those options aren't there for people like
> you who get annoyed when the package manager tries to stop them doing
> something stupid. In fact, probably it would be better if they weren't
> there at all.

> I suggest you learn to trust RPM. It is far more likely to be right than
> you are, as you have already demonstrated. Do yourself a favour and
> forget you ever heard of --force and --nodeps.

Keep in mind that RPMs are only as good the person who's putting them
together. I don't mean to imply anything here, but I bring this up as a
matter of point. Would you trust an RPM from someone you don't know?
It's up to the packager to make sure the "Requires" and "Provides" are
accurate and match what other packages expect. It's also up to them to
ensure that there are no conflicting files. For a site as large as
atrpms, Axel would also have to make sure that all the dependencies are
provided as RPMs or otherwise satisfied by the target distribution.

For my MythTV install, I tried to stay with RPMs but I needed a version
of the ivtv driver that was not yet packaged. Yes, I could have created
my own RPM as I do for packages I use frequently, but with only one Myth
box and no guarantee that an upgraded driver would work, a tarball
looked pretty good. This was not the cause of the dependency problem
though since nothing requires ivtv except itself.

I also went for the newer ivtv-firmware rpm (before realizing I needed
the new driver), this rpm had both the encoder and decoder firmware.
Since I was getting this from Axel's site I figured it would work (see,
I do trust RPMS) sadly it didn't. That appears to have changed by now
though. I see posts from people who have claimed to install both. A feat
I was not willing to do, and one that would require a --force.

What I may have not made clear is that the real problems had occurred
before I went around RPM. Python4 was not selected for installation even
though some RPMs required it. Also a package from atrpms conflicted with
the yum RPM, the version installed by Fedora.

Don't get me wrong, I'm not against RPMS. I use RPMs every day on lots
of servers and workstations and I rely on it to distribute my software.
When it becomes more burden that it's worth I circumvent it. Part of
this is laziness and part of it is because I make my living as an
administrator. My customers won't see the value of spending their money
to create RPMs.

> FWIW, I have used Axel's RPMs for myth 0.15, 0.16, and currently the CVS
> RPMs. On FC2, now FC3. I've never had a single dependency problem. But
> then I've never used --nodeps either.
> 

-- 
John Van Ostrand <john at netdirect.ca>
Net Direct Inc.



More information about the mythtv-users mailing list