Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > linux.debian.user > #207442 > unrolled thread

Exploring package interrelationships

Started byRichard Owlett <rowlett@cloud85.net>
First post2019-04-14 14:40 +0200
Last post2019-04-15 18:40 +0200
Articles 14 — 8 participants

Back to article view | Back to linux.debian.user


Contents

  Exploring package interrelationships Richard Owlett <rowlett@cloud85.net> - 2019-04-14 14:40 +0200
    Re: Exploring package interrelationships Nicholas Geovanis <nickgeovanis@gmail.com> - 2019-04-14 15:00 +0200
      Re: Exploring package interrelationships Richard Owlett <rowlett@cloud85.net> - 2019-04-14 15:40 +0200
        Re: Exploring package interrelationships <tomas@tuxteam.de> - 2019-04-15 09:20 +0200
          Re: Exploring package interrelationships Richard Owlett <rowlett@cloud85.net> - 2019-04-15 13:10 +0200
          Re: Exploring package interrelationships Cindy Sue Causey <butterflybytes@gmail.com> - 2019-04-15 15:30 +0200
            Re: Exploring package interrelationships Gene Heskett <gheskett@shentel.net> - 2019-04-15 16:10 +0200
    Re: Exploring package interrelationships David Wright <deblis@lionunicorn.co.uk> - 2019-04-14 16:40 +0200
    Re: Exploring package interrelationships Peter Wiersig <peter@friesenpeter.de> - 2019-04-15 10:40 +0200
      Re: Exploring package interrelationships Richard Owlett <rowlett@cloud85.net> - 2019-04-15 16:00 +0200
        Re: Exploring package interrelationships David Wright <deblis@lionunicorn.co.uk> - 2019-04-15 16:20 +0200
          Re: Exploring package interrelationships Richard Owlett <rowlett@cloud85.net> - 2019-04-15 17:20 +0200
            Re: Exploring package interrelationships David Wright <deblis@lionunicorn.co.uk> - 2019-04-15 18:00 +0200
            Re: Exploring package interrelationships Curt <curty@free.fr> - 2019-04-15 18:40 +0200

#207442 — Exploring package interrelationships

FromRichard Owlett <rowlett@cloud85.net>
Date2019-04-14 14:40 +0200
SubjectExploring package interrelationships
Message-ID<xMMAq-6ME-7@gated-at.bofh.it>
Long term goal: *personal* definition of a minimalist Debian

current goal: grok how packages interact

current *test case*/example: a very minimal install of MATE
An illustration of the opposite of what I want is task-mate-desktop.
It's description states:
> This task package is used to install the Debian desktop, featuring
> the MATE desktop environment, and with other packages that Debian
> users expect to have available on the desktop.In what I want only the top level menu headings [Applications Places 
System] would exist. Under Applications the sub-headings [Accessories 
Education Graphics etc] may exist but their contents would be empty.

proposed procedure:
Do a minimal install of a CLI system.
Follow with "apt-get install" of a very minimal set of MATE components.
[Pluma, Caja, and Synaptic would be included]

Under "https://packages.debian.org/stretch/" I have been exploring 
entries under task-mate-desktop, mate-desktop-environment, 
mate-desktop-environment-core, and desktop-base.

I'm do getting a good visualization of how to reach my goal. Something 
that acted on repositories as apt-cache does on the current machine's 
cache would be useful.

In past conversations it has been suggested that I do a typical install 
and just un-install the un-desired elements:

That is undesirable for two primary reasons:
1. I would not reach my primary goal of "grok how packages interact".
2. Uncertainty of what the resulting "thing" would be.
    [While channel surfing recently, I caught a visual example. A
     children's show was exploring mixing and sorting. The 1st example
     was putting colored balls in a glass bowl. There was no problem
     separating the balls into 2 sets. The 2nd example was mixing two
     glasses of water with red dye in one and green in the other. The
     mixing could not be undone.]

Comments/suggestions/readings/search terms.

TIA

[toc] | [next] | [standalone]


#207443

FromNicholas Geovanis <nickgeovanis@gmail.com>
Date2019-04-14 15:00 +0200
Message-ID<xMMTL-6Tx-3@gated-at.bofh.it>
In reply to#207442

[Multipart message — attachments visible in raw view] — view raw

On Sun, Apr 14, 2019 at 7:33 AM Richard Owlett <rowlett@cloud85.net> wrote:

>
> That is undesirable for two primary reasons:
> 1. I would not reach my primary goal of "grok how packages interact".
>

I think a better way to proceed, which will be quite confusing itself :-)
is to
simply explore those relationships using the tools that there are. Rather
than
exploring it less-systematically under trial by fire. If you want to
experiment live
while exploring, fire up a VM at the same time.

2. Uncertainty of what the resulting "thing" would be.
>

This shouldn't be an issue IMO. Why is that important?
It's a Debian system with certain packages installed. If you work
in a data center and manage hundreds or thousands of servers,
for example, your config management software cares. You don't.
But I see no reason to care or stress on this in any context.


> Comments/suggestions/readings/search terms.
>
> TIA
>
>
>
>
>

[toc] | [prev] | [next] | [standalone]


#207446

FromRichard Owlett <rowlett@cloud85.net>
Date2019-04-14 15:40 +0200
Message-ID<xMNwt-7mU-1@gated-at.bofh.it>
In reply to#207443
On 04/14/2019 07:50 AM, Nicholas Geovanis wrote:
> On Sun, Apr 14, 2019 at 7:33 AM Richard Owlett <rowlett@cloud85.net 
> <mailto:rowlett@cloud85.net>> wrote:
> 
> 
>     That ...

Moderate trimming is good.  Yours destroyed the context of "That".
It referred to:
>>> In past conversations it has been suggested that I do a typical
>>> install and just un-install the un-desired elements: 

>> That is undesirable for two primary reasons:
>>     1. I would not reach my primary goal of "grok how packages interact".
> 
> 
> I think a better way to proceed, which will be quite confusing itself 
> :-) is to
> simply explore those relationships using the tools that there are. 

What tools are you referring to?

> Rather than
> exploring it less-systematically under trial by fire. If you want to 
> experiment live
> while exploring, fire up a VM at the same time.
> 
>     2. Uncertainty of what the resulting "thing" would be.
> 
> 
> This shouldn't be an issue IMO. Why is that important?

I've been down this route before. It gave an unsatisfactory 
approximation of what was desired. I was explicitly specifying an 
defective approach that I am not interested in following.
YMMV ;/

> It's a Debian system with certain packages installed. If you work
> in a data center and manage hundreds or thousands of servers,
> for example, your config management software cares. You don't.
> But I see no reason to care or stress on this in any context.
> 
>     Comments/suggestions/readings/search terms.
> 
>     TIA

[toc] | [prev] | [next] | [standalone]


#207455

From<tomas@tuxteam.de>
Date2019-04-15 09:20 +0200
Message-ID<xN44h-KC-5@gated-at.bofh.it>
In reply to#207446

[Multipart message — attachments visible in raw view] — view raw

On Sun, Apr 14, 2019 at 08:38:48AM -0500, Richard Owlett wrote:

[...]

> >>>In past conversations it has been suggested that I do a typical
> >>>install and just un-install the un-desired elements:

Start with what debootstrap [1] gives you and work up from there.
This is what any regular Debian installer does to get itself started.
That's what I do when trying to get a minimal install.

But beware: debootstrap is perhaps more minimal than what you might
consider minimal. No GUI. Not even an SSH daemon. But apt.

Debootstrap is a harsh mistress (yeah, stolen from somewhere). But
you might end up liking it once you get the hang of it...

Cheers

[1] Apart from having (of course) its extensive manual page,
  it has an entry in the Debian wiki:
  https://wiki.debian.org/Debootstrap

-- t

[toc] | [prev] | [next] | [standalone]


#207476

FromRichard Owlett <rowlett@cloud85.net>
Date2019-04-15 13:10 +0200
Message-ID<xN7ES-2Yr-11@gated-at.bofh.it>
In reply to#207455
On 04/15/2019 02:12 AM, tomas@tuxteam.de wrote:
> On Sun, Apr 14, 2019 at 08:38:48AM -0500, Richard Owlett wrote:
> 
> [...]
> 
>>>>> In past conversations it has been suggested that I do a typical
>>>>> install and just un-install the un-desired elements:
> 
> Start with what debootstrap [1] gives you and work up from there.
> This is what any regular Debian installer does to get itself started.
> That's what I do when trying to get a minimal install.

Brings back memories. I had tried to do something similar years ago when 
I first got started with Linux. I didn't have enough background and got 
thoroughly *lost*. THANKS for the reminder. IIRC Multistrap may match my 
mindset better. I've a lot of re-reading to do (jessie was current last 
time ;)

> 
> But beware: debootstrap is perhaps more minimal than what you might
> consider minimal. No GUI. Not even an SSH daemon. But apt.

ROFL I once said there was no need for floppies larger than 1.44M - does 
that date me?

> 
> Debootstrap is a harsh mistress (yeah, stolen from somewhere). But
> you might end up liking it once you get the hang of it...

I liked it bu NEVER did get the hang of it ;)


> 
> [1] Apart from having (of course) its extensive manual page,
>    it has an entry in the Debian wiki:
>    https://wiki.debian.org/Debootstrap
> 
> -- t
> 

[toc] | [prev] | [next] | [standalone]


#207490

FromCindy Sue Causey <butterflybytes@gmail.com>
Date2019-04-15 15:30 +0200
Message-ID<xN9Qm-4eK-11@gated-at.bofh.it>
In reply to#207455
On 4/15/19, tomas@tuxteam.de <tomas@tuxteam.de> wrote:
>
> Debootstrap is a harsh mistress (yeah, stolen from somewhere). But
> you might end up liking it once you get the hang of it...


Don't know if I ever heard that, at least not that I can remember that
topic specifically. I just know that debootstrap always *WORKS FOR
ME*.

Well, *IT WORKS* except for that one time that root kept yelling, "I
HAVE NO NAME!" That eventually turned out to be caused by symlinking
/debootstrap-directory/var/cache/apt/archives to an offline dotDEB
file hoard. Accidentally discovered that using "mount -B" instead of
"ln -s" solved root's identity crisis.

I've been "bored" lately and so was going to spend some time nosing
around to see if any other methods are dialup friendly. What you
said... I'll have to poke around at that, too. Something in the very
back of my head is nagging that I may have read some disgruntled
chatter LONG before it ever made enough sense to find a more permanent
spot toward the front of the class where its topic would remain more..
[conscionable]. :)

Cindy :)
-- 
Cindy-Sue Causey
Talking Rock, Pickens County, Georgia, USA

* runs with birdseed *

[toc] | [prev] | [next] | [standalone]


#207493

FromGene Heskett <gheskett@shentel.net>
Date2019-04-15 16:10 +0200
Message-ID<xNat4-4HM-9@gated-at.bofh.it>
In reply to#207490
On Monday 15 April 2019 09:28:12 Cindy Sue Causey wrote:

> On 4/15/19, tomas@tuxteam.de <tomas@tuxteam.de> wrote:
> > Debootstrap is a harsh mistress (yeah, stolen from somewhere). But
> > you might end up liking it once you get the hang of it...
>
Somewhat off topic, but your reading material in your formative years 
Cindy, is sadly lacking in familiarity with the output of one of the 
last century's most prolific  Sci Fi authors named Robert A. Heinlein. 
Specifically his treatise on living on or "in" the moon.  Go to your 
public library and check out "The moon is a harsh mistress" and remedy 
that lack. ;-)  I think you'll go back and get some more of his output.

> Don't know if I ever heard that, at least not that I can remember that
> topic specifically. I just know that debootstrap always *WORKS FOR
> ME*.
>
> Well, *IT WORKS* except for that one time that root kept yelling, "I
> HAVE NO NAME!" That eventually turned out to be caused by symlinking
> /debootstrap-directory/var/cache/apt/archives to an offline dotDEB
> file hoard. Accidentally discovered that using "mount -B" instead of
> "ln -s" solved root's identity crisis.
>
> I've been "bored" lately and so was going to spend some time nosing
> around to see if any other methods are dialup friendly. What you
> said... I'll have to poke around at that, too. Something in the very
> back of my head is nagging that I may have read some disgruntled
> chatter LONG before it ever made enough sense to find a more permanent
> spot toward the front of the class where its topic would remain more..
> [conscionable]. :)
>
> Cindy :)

I had forgotten there was still diakup country Cindy, my sympathies. But 
with Pi at the FCC, its not going to be fixed on his watch, he is in big 
telecom's pocket. "A Republic, Mam, if we can keep it", but we've not 
been doing that good a job.

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page <http://geneslinuxbox.net:6309/gene>

[toc] | [prev] | [next] | [standalone]


#207447

FromDavid Wright <deblis@lionunicorn.co.uk>
Date2019-04-14 16:40 +0200
Message-ID<xMOsy-7WD-17@gated-at.bofh.it>
In reply to#207442
On Sun 14 Apr 2019 at 07:32:36 (-0500), Richard Owlett wrote:
> Long term goal: *personal* definition of a minimalist Debian

Debian is in no way minimilist: in fact, it's the opposite,
"The universal operating system" claiming over 51000 packages.

You're the one claiming to be a minimalist, though we may have our
doubts because your "minimal CLI" system is to include tools like
synaptic which requires a graphical system of one kind or another.

But by having a *personal* definition (your emphasis) of what's
minimalist^H^H^H, you've reserved the role of judge and jury on
any suggestions made here. As usual.

> current goal: grok how packages interact

AIUI, which is but partially, the interactions at the granularity
of installation (ie ignoring library calls and suchlike) are
encapsulated in the Packages file's {Pre-,}Depends: lines.

I've taken a look at these using Python dictionaries, but it would
appear that they actually form a type of directed graph, for which
mathematicians have a battery of tools, I'm sure. (It's not merely
"simple" because two packages can depend on each other, but I haven't
looked for loops or other complications.)

> current *test case*/example: a very minimal install of MATE
> An illustration of the opposite of what I want is task-mate-desktop.
> It's description states:
> > This task package is used to install the Debian desktop, featuring
> > the MATE desktop environment, and with other packages that Debian
> > users expect to have available on the desktop.In what I want only
> > the top level menu headings [Applications Places
> System] would exist. Under Applications the sub-headings [Accessories
> Education Graphics etc] may exist but their contents would be empty.
> 
> proposed procedure:
> Do a minimal install of a CLI system.
> Follow with "apt-get install" of a very minimal set of MATE components.
> [Pluma, Caja, and Synaptic would be included]
> 
> Under "https://packages.debian.org/stretch/" I have been exploring
> entries under task-mate-desktop, mate-desktop-environment,
> mate-desktop-environment-core, and desktop-base.
> 
> I'm do getting a good visualization of how to reach my goal. Something
> that acted on repositories as apt-cache does on the current machine's
> cache would be useful.

Well, there's your answer then. Study how it works. Download the
source and analyse it.

> In past conversations it has been suggested that I do a typical
> install and just un-install the un-desired elements:
> 
> That is undesirable for two primary reasons:
> 1. I would not reach my primary goal of "grok how packages interact".
> 2. Uncertainty of what the resulting "thing" would be.
>    [While channel surfing recently, I caught a visual example. A
>     children's show was exploring mixing and sorting. The 1st example
>     was putting colored balls in a glass bowl. There was no problem
>     separating the balls into 2 sets. The 2nd example was mixing two
>     glasses of water with red dye in one and green in the other. The
>     mixing could not be undone.]

I don't understand. You're implying that the install-then-uninstall
method will mix packages like dye in water, and I've seen no evidence
that Debian systems behave like that (though I can't speak for synaptic).

> Comments/suggestions/readings/search terms.

You've answered it here: it's in the package resolvers. At this stage
you might be expected to have an idea of how many distinct resolvers
there are in Debian's various installers, apt-get, aptitude, dselect,
etc. Myself—I don't. I've always found apt-get's satisfactory enough
without investigating further.

Cheers,
David.

[toc] | [prev] | [next] | [standalone]


#207466

FromPeter Wiersig <peter@friesenpeter.de>
Date2019-04-15 10:40 +0200
Message-ID<xN5jH-1pS-3@gated-at.bofh.it>
In reply to#207442
Richard Owlett <rowlett@cloud85.net> writes:
> Long term goal: *personal* definition of a minimalist Debian
>
> current goal: grok how packages interact

http://www.macfreek.nl/memory/Dependency_Graph_Debian_Packages
TLDR:
$ apt-cache dotty mate-desktop > dependency-graph.dot
$ dot -Tpng  dependency-graph.dot

That either creates files named after the input or at least shows a
graphic of the package you specified.

I think I remember and older planet.debian.org post where someone else
had done similar.  Have not found it in a very short research session.


Alternatively explore interactive with aptitude when you disable the
solver there and pick the resulting installation by hand, and decide
which suggest/recommend you follow, and which non-essential package you
might even not install.

Peter

[toc] | [prev] | [next] | [standalone]


#207492

FromRichard Owlett <rowlett@cloud85.net>
Date2019-04-15 16:00 +0200
Message-ID<xNajo-4oX-7@gated-at.bofh.it>
In reply to#207466
On 04/15/2019 03:39 AM, Peter Wiersig wrote:
> Richard Owlett <rowlett@cloud85.net> writes:
>> Long term goal: *personal* definition of a minimalist Debian
>>
>> current goal: grok how packages interact
> 
> http://www.macfreek.nl/memory/Dependency_Graph_Debian_Packages
> TLDR:
> $ apt-cache dotty mate-desktop > dependency-graph.dot
> $ dot -Tpng  dependency-graph.dot


After I posted I found a reference using Graphiz with "apt-cache dotty"
I haven't read the documentation yet.

I tried you example and got a command not found at second line.
I found a man page for "dot" then went looking for a package to provide 
it -- found/installed xdot.

Using "dot -Tps  dependency-graph.dot > owl.ps" I got a displayable, if 
not readable, graph. There was a warning in what I had read about 
"dotty" about viewablity issues if the graph was too complex. Suspect 
that is problem.



> 
> That either creates files named after the input or at least shows a
> graphic of the package you specified.
> 
> I think I remember and older planet.debian.org post where someone else
> had done similar.  Have not found it in a very short research session.
> 
> 
> Alternatively explore interactive with aptitude when you disable the
> solver there and pick the resulting installation by hand, and decide
> which suggest/recommend you follow, and which non-essential package you
> might even not install.

I can't parse that sentence. By context I suspect I can do some 
searching which will clarify.

> 
> Peter
> 
> 

[toc] | [prev] | [next] | [standalone]


#207494

FromDavid Wright <deblis@lionunicorn.co.uk>
Date2019-04-15 16:20 +0200
Message-ID<xNaCJ-4L2-3@gated-at.bofh.it>
In reply to#207492
On Mon 15 Apr 2019 at 08:50:12 (-0500), Richard Owlett wrote:
> On 04/15/2019 03:39 AM, Peter Wiersig wrote:
> > Richard Owlett <rowlett@cloud85.net> writes:
> > > Long term goal: *personal* definition of a minimalist Debian
> > > 
> > > current goal: grok how packages interact
> > 
> > http://www.macfreek.nl/memory/Dependency_Graph_Debian_Packages
> > TLDR:
> > $ apt-cache dotty mate-desktop > dependency-graph.dot
> > $ dot -Tpng  dependency-graph.dot
> 
> After I posted I found a reference using Graphiz with "apt-cache dotty"
> I haven't read the documentation yet.
> 
> I tried you example and got a command not found at second line.
> I found a man page for "dot" then went looking for a package to
> provide it -- found/installed xdot.

I'm not sure we need the blow by blow account.

> Using "dot -Tps  dependency-graph.dot > owl.ps" I got a displayable,
> if not readable, graph. There was a warning in what I had read about
> "dotty" about viewablity issues if the graph was too complex. Suspect
> that is problem.

Add   -o APT::Cache::GivenOnly=1   to the commandline, and you'll get
a much smaller graph. I haven't figured out why python and
python-requests are missng from the graph, or is this a stretch/buster
difference. (I'm comparing the graph with the Packages entry:

Package: mate-desktop
Version: 1.16.2-2
Architecture: amd64
Replaces: mate-desktop-gnome
Depends: hicolor-icon-theme, libmate-desktop-2-17 (>= 1.10.0),
 mate-desktop-common (= 1.16.2-2), python, python-requests, libatk1.0-0
 (>= 1.12.4), libc6 (>= 2.4), libcairo-gobject2 (>= 1.10.0), libcairo2
 (>= 1.2.4), libgdk-pixbuf2.0-0 (>= 2.22.0), libglib2.0-0 (>= 2.28.0),
 libgtk-3-0 (>= 3.0.0), libpango-1.0-0 (>= 1.14.0), libpangocairo-1.0-0
 (>= 1.14.0), libstartup-notification0 (>= 0.2), libxrandr2
Recommends: mate-user-guide
Breaks: mate-desktop-gnome

[edited].)

> > That either creates files named after the input or at least shows a
> > graphic of the package you specified.
> > 
> > I think I remember and older planet.debian.org post where someone else
> > had done similar.  Have not found it in a very short research session.
> > 
> > 
> > Alternatively explore interactive with aptitude when you disable the
> > solver there and pick the resulting installation by hand, and decide
> > which suggest/recommend you follow, and which non-essential package you
> > might even not install.
> 
> I can't parse that sentence. By context I suspect I can do some
> searching which will clarify.

Yes, it's often difficult to follow explanations of GUI processes
because they're interactive by nature. Sometimes a video is more
help than written instructions. The modern world …

Cheers,
David.

[toc] | [prev] | [next] | [standalone]


#207505

FromRichard Owlett <rowlett@cloud85.net>
Date2019-04-15 17:20 +0200
Message-ID<xNbyN-5kX-1@gated-at.bofh.it>
In reply to#207494
On 04/15/2019 09:15 AM, David Wright wrote:
> [snip]
> 
>> Using "dot -Tps  dependency-graph.dot > owl.ps" I got a displayable,
>> if not readable, graph. There was a warning in what I had read about
>> "dotty" about viewablity issues if the graph was too complex. Suspect
>> that is problem.
> 
> Add   -o APT::Cache::GivenOnly=1   to the commandline, and you'll get
> a much smaller graph. I haven't figured out why python and
> python-requests are missng from the graph, or is this a stretch/buster
> difference. (I'm comparing the graph with the Packages entry:
> 

Neither
   > $ dot -Tps -o APT::Cache::GivenOnly=1 dependency-graph.dot > owl2.ps
Nor
   > $ dot -o APT::Cache::GivenOnly=1 dependency-graph.dot
gave useful output.

I'll do some reading and try on something simpler.
Thanks

[toc] | [prev] | [next] | [standalone]


#207508

FromDavid Wright <deblis@lionunicorn.co.uk>
Date2019-04-15 18:00 +0200
Message-ID<xNcbv-5yw-1@gated-at.bofh.it>
In reply to#207505
On Mon 15 Apr 2019 at 10:13:31 (-0500), Richard Owlett wrote:
> On 04/15/2019 09:15 AM, David Wright wrote:
> > [snip]
> > 
> > > Using "dot -Tps  dependency-graph.dot > owl.ps" I got a displayable,
> > > if not readable, graph. There was a warning in what I had read about
> > > "dotty" about viewablity issues if the graph was too complex. Suspect
> > > that is problem.
> > 
> > Add   -o APT::Cache::GivenOnly=1   to the commandline, and you'll get
> > a much smaller graph. I haven't figured out why python and
> > python-requests are missng from the graph, or is this a stretch/buster
> > difference. (I'm comparing the graph with the Packages entry:
> > 
> 
> Neither
>   > $ dot -Tps -o APT::Cache::GivenOnly=1 dependency-graph.dot > owl2.ps
> Nor
>   > $ dot -o APT::Cache::GivenOnly=1 dependency-graph.dot
> gave useful output.

I'm not sure how you expect to progress if you expect people here,
following your blow by blow account, to write each commandline for
you in full.

It's fairly obvious that APT::Cache::GivenOnly applies to the
apt-cache command, given the words it has in common. In fact,
it's likely you have some options like this in your /etc/apt/
tree, eg:

/etc/apt/apt.conf.d/00trustcdrom :
APT::Authentication::TrustCDROM "true";

Cheers,
David.

[toc] | [prev] | [next] | [standalone]


#207513

FromCurt <curty@free.fr>
Date2019-04-15 18:40 +0200
Message-ID<xNcOe-61X-13@gated-at.bofh.it>
In reply to#207505
On 2019-04-15, Richard Owlett <rowlett@cloud85.net> wrote:
>> 
>
> Neither
>    > $ dot -Tps -o APT::Cache::GivenOnly=1 dependency-graph.dot > owl2.ps
> Nor
>    > $ dot -o APT::Cache::GivenOnly=1 dependency-graph.dot
> gave useful output.
>
> I'll do some reading and try on something simpler.
> Thanks
>
>

That doesn't seem to be how you're supposed to do it.

I'm reading

Create the graph (with no package given all packages in the cache are
graphed):

 apt-cache dotty apache2 > apache-dependency-graph.dot

Visualize the graph:

 dot -Tpng apache-dependency-grap.dot

But there's still too much stuff in there maybe to be legible. DW
suggests a restriction of the graph to only the packages given on the
command line. 

 apt-cache -o APT::Cache::GivenOnly=1 dotty apache2 apache2-common (etc.).

http://www.macfreek.nl/memory/Dependency_Graph_Debian_Packages

[toc] | [prev] | [standalone]


Back to top | Article view | linux.debian.user


csiph-web