Mailing List Archive

2rpm or not 2rpm, that is the question (was: Updated MythTV RPMs)
Cedar,

why so hostile with rpms? They don't bite ... ;)

On Fri, Feb 14, 2003 at 05:28:13PM -0800, Cedar McKay wrote:
> why would you want to use the rpms in this situation? You are just adding
> another layer of complexity and potential screwups to an already complicated
> install process. [...]
> [...] Just follow along, and when you get to mythtv it is a simple configure,
> make, make install. [...]

Now is setting up mythtv "a complicated process" or a "simple configure, make,
make install" (rhetorical)?

You should ask yourself the purpose of rpms (or of Debian dpkg or any
packaging system). And not only for mythtv, but also for anything else: They
should ease the installation and setup of software and deal with
dependencies. They don't suggest replacing documentation!

Especially multimedia applications like mythtv and mplayer are targeted (also)
for average users, that should not have to be software gurus or be able to
follow a rather technical setup and possibly deal with problems that are not
covered.

rpms/dpkgs do raise the acceptance and usage of any software bit.

> In my opinion there are too many things that rpms (probably) don't take into
> account here.

Instead of guessing and spreading FUD, you should check - you are running
RH8.0 yourself, so it would only have taken a minute ...

As you know everything is technically possible in the framework of rpms/dpkgs,
so it is not a question about packaging or not, it is a question of their
quality.

Don't drive people away from the packaging concept. Try instead to be
constructive. If you do find true shortcomings in the packages you should
suggest improvements.

> For instance, does your rpm know that libexpat is broken in RH8?

Yes. It will whine about the broken expat and require you or any automated
packaging framework to install the working version. In the automated variant
(e.g. apt-get) it will even suggest to get it for you and install it. You will
only have to approve.

Surprised? Yes, packaging managers do have some intelligence ...

> Does your rpm set the LANG variable to "us"?

No, you know it should not (although it could)!

> Or know that sometimes you have to run ldconfig multiple times?

Yes, again ...

> Or that sometimes you do need to recompile the kernel to get lirc running
> and sometimes you don't.

You must be kidding.

Your denial against packaging software is remarkable. Next thing you will
suggest is reformating our hard disks and go Linux From Scratch.

> And then you won't have to tear _your_ hair out bumbling around with rpms
> made by who knows who, and inventing new problems that nobody really wants
> to solve.
>
> ahhhh that felt good

This sounds too personal for me to go on.

If you feel like the rpms needing improvement your comments are welcome. I
am really not fond of flaming, ranting and politics.

> cedar
--
Axel.Thimm@physik.fu-berlin.de
Re: 2rpm or not 2rpm, that is the question (was: Updated MythTV RPMs) [ In reply to ]
> why so hostile with rpms? They don't bite ... ;)

I'm not. The documentation already suggests rpms for every required
prerequisite that Red Hat happens to offer as an rpm. I _am_ nervous
about unofficial, unsupported rpms like the mythtv rpm and the alsa
rpm. If rpms are the way to go, why doesn't alsa-project.org offer an
rpm itself? In general I think it is the best idea to install in the
manner intended by the creators of the software. Any time a project
itself offers an rpm I'm open to using it.


> Now is setting up mythtv "a complicated process" or a "simple
> configure, make,
> make install" (rhetorical)?
>
The "complicated process" is getting the prerequisites installed
(particularly xmltv) and fighting through the libexpat, lang=c and so
on. The actual installation of mythtv is dirt simple whether you use an
rpm or configure, make, make install.


> You should ask yourself the purpose of rpms (or of Debian dpkg or any
> packaging system).
> And not only for mythtv, but also for anything else: They
> should ease the installation and setup of software and deal with
> dependencies.
The threshold for getting mythtv installed (at least on Red Hat) right
now is high. Much higher than I would like. But really right now the
hardest part of the install is probably xmltv, and lirc. Maybe sound
too. I don't see the mythtv rpm helping with the hard parts, but they
do cause newbies to depart from the flow of the documentation, and
strike off into unknown/unsupported territory.


> They don't suggest replacing documentation!
>
The documentation does not describe how to deal with installing using
rpms. So right now, newbies come along and they are caught between
following the documentation, or taking a shortcut and using the "easy"
rpm, which at this point is undocumented and unknown to most of the
people prepared to help newbies with their problems. I would be _way_
more receptive to the rpm package if it had also been supplied with all
the needed modifications to the current documentation, so a newbie
could just follow along with a document (using rpms or not) and not
have to make any big leaps into unknown territory.


> Especially multimedia applications like mythtv and mplayer are
> targeted (also)
> for average users, that should not have to be software gurus or be
> able to
> follow a rather technical setup and possibly deal with problems that
> are not
> covered.
Mythtv may be appealing to non-gurus, but it is still a pretty tough
install. Being appealing to a certain type, does not make it
_appropriate_ for that same type. That said, that is why mythtv has
some of the most complete documentation I've seen in a linux based
project. Someone with the right hardware and a few red hat disks could
follow along with the documentation, and as long as they read
carefully, they would never lose their way, never have to strike into
unknown territory, and not have to know too much.


> rpms/dpkgs do raise the acceptance and usage of any software bit.
>
That might become desirable with the 1.0 release, you would have to ask
isaac.




>
> Don't drive people away from the packaging concept. Try instead to be
> constructive. If you do find true shortcomings in the packages you
> should
> suggest improvements.

OK, here is a shortcoming: they don't come with comprehensive
documentation.


>> Does your rpm set the LANG variable to "us"?
> No, you know it should not (although it could)!
maybe it should ask whether to change it for the system or for just the
current user. There is another constructive suggestion.


>> Or know that sometimes you have to run ldconfig multiple times?
>
> Yes, again ...
>
really? It can? I'm impressed.


>> Or that sometimes you do need to recompile the kernel to get lirc
>> running
>> and sometimes you don't.
>
> You must be kidding.

kinda : )



> Your denial against packaging software is remarkable. Next thing you
> will
> suggest is reformating our hard disks and go Linux From Scratch.

I was funny to see you get more and more worked up as you went along.


> If you feel like the rpms needing improvement your comments are
> welcome. I
> am really not fond of flaming, ranting and politics.
easy big fella.



Here is the bottom line. If you (or anyone) submit documentation to go
along with the rpms that is as good and complete as the current
documentation, and the rpms can be verified as high quality, and Isaac
and Robert approve, then I would welcome rpms. If they proved easier
than the regular install process I would support getting rid of the
configure, make, make install method.



cheers,


cedar
Re: 2rpm or not 2rpm, that is the question (was: Updated MythTV RPMs) [ In reply to ]
On Saturday 15 February 2003 06:02 am, Axel Thimm wrote:
> They
> should ease the installation and setup of software and deal with
> dependencies.

They also allow software management through the RPM database and keep track of
modified and changed files when installing/upgrading.

--
Hoyt
http://www.maximumhoyt.com

Run with scissors. Remove mattress tags. Top post. Be a rebel.