R SIG Debian - Aug 2023 - Is r2u at 3.4.1? [branch about handling package collisions under Ubuntu/Debian]

Hi Chris,

Thanks for posting here, and sorry I missed it for two days due to sorting
mailing list traffic into folders and clearly not glancing at all of them.

On 13 August 2023 at 09:12, Chris Evans wrote:
| I am putting this here as it may be of general and not just my own 
| interest.? I am currently running 4.3.1 on Ubuntu 22.04.2 LTS but am 
| hitting an issue that the magick package won't update because 
| libmagick++-dev won't update because, presumably, of two repositories in 
| my sources.list disagreeing about versions required.? That's not so

That can happen, and pinning can help. I would suggest to look at 'apt-cache
polict nameofthepackagehere'. (See below for concrete example.)

(Also: how come there are two libmagick++-dev, ie why do you have a 2nd one,
ie do you know which additional repo you turned on?)

| urgent an issue for me that I wont just wait to see if it sorts itself 
| out.? In my experience such collisions sometimes do.
| 
| However, it prompted me to some questions:
| 
| 1) Is this telling me it's time I went to r2u which does seem a "very
| good thing"?? Any reason NOT to change?

Hm, I don't exactly follow.

r2u is a good thing. I use it more and more, so do others. Traffic is up, and
we get more and more use cases. Works really well on 22.04 LTS and later. (I
happen to run 23.04 now.)

But r2u cannot prevent you from encountering _other conflicts from other
repos_ as you may have here with magick++. (It may of course reduce the need
for other repos, and that would help.)
 
| 2) I read https://eddelbuettel.github.io/r2u/, not for the first time, 
| and it looks simple, even I should? be able to do that.? Any gotchas?? 
| See #3:
| 
| 3) but it says things I don't really understand about the R version:
|  ?? "Current versions are based on R 4.3.0, and BioConductor release 
| 3.17 packages are provided when required by CRAN packages. Binaries are 
| still R 4.2.* based (unless a forced rebuild was required) but the 
| containers provide R 4.3.0. We expect to switch to R 4.3.0-based builds 
| very soon."

That sounds like an old paragraph I should update. We are now at R 4.3.1, of
course, and BioConductor 3.17 (released right after R 4.3.0). The next
sentence is from the 4.2 -> 4.3 transition, r2u has looong switched to 4.3
generation too so I will delete this.

Thanks for the heads-up.
 
|  ?? 3a) I don't understand why "binaries", presumably binary
packages,
| would be at a different release from ... what?? What is not binary?? 
| Pure R source packages?

R binaries from CRAN per the usual README at
   https://cloud.r-project.org/bin/linux/ubuntu/
which we pointed to for what must be well over 10 years. (That README still
only talks c2d4u, which is of course still up and well, but may be time to
add a link to r2u too.)

|  ?? 3b) _IS_ it still at 4.3.0 as I'm at 3.4.1 and would be a bit 
| reluctant to go backwards down the release sequence?

No nothing is at R 4.3.0.

You can always check but adding the apt entry to your sources.list file and
doing 'apt-cache policy r-base-core' to see which package version is /
would
be pulled it.

| TIA and enormous respect to Dirk for this and all the other work he does 
| for R,

Thanks for those very kinds words. Now please help me to make the READMEs
clearer. I will start by nixing the obsolete sentence you alerted me to, so
thans for that.

Cheers, Dirk

| Visiting Professor, UDLA, Quito, Ecuador & Honorary Professor, 
| University of Roehampton, London, UK.

Nice. I got to workshop-keynote once in Guayaquil and regret not making over
to Quito.

-- 
dirk.eddelbuettel.com | @eddelbuettel | edd at debian.org

Chris,

I edited the README.md for r2u (also the main page at
https://eddelbuettel.github.io/r2u) in a few places, that should hopefully
address a few of the snafus you found.

As for your question 'is r2u running R version x.y.z' there are few way
to
find out

 - if you have docker:  docker run --rm -ti rocker/r2u
   with the default 'latest' tag for 22.04, tags 22.04 and 20.04 also
work

 - in the webbrowser (!!) as long demonstrated on that page via gitpad.io

 - in the webbrowser (!!) or via the 'code' editor as added last weekend
in
   a new vignette https://eddelbuettel.github.io/r2u/vignettes/Codespaces/
   (also on my blog, and tweeted & tooted about).

That last part is compelling as GitHub gives a number of free compute hours
for codespace on either a 2-core, 4gb instance or a 4-core, 16gb instance
which is really not too shabby. (Instead of clicking on '+' select
'...' and
'new with options', pick bigger instance and/or other options).

Cheers, Dirk

-- 
dirk.eddelbuettel.com | @eddelbuettel | edd at debian.org

This is definitely tangential to the list: I'm on Ubuntu (22.04.2 LTS)
not Debian and I'm sure this is about issues in the Ubuntu package 
management
on my machine, R is only revealing them.

The original subject line came from me wondering if my going over to the r2u
repository would solve the problem.? However, as I think Dirk said, that's
only really likely to be answered by trying it and I would rather see if I
can understand what has gone wrong before I do anything radical.

I am also hoping that having this, partially OTT thread on the list may
help as I doubt if I'll be the last to hit this sort of package management
nightmare.? Some of the issues clearly are pure OS issues but the impact
for me is definitely on R.

On 16/08/2023 00:06, Dirk Eddelbuettel wrote:
> Hi Chris,
>
> Thanks for posting here, and sorry I missed it for two days due to sorting
> mailing list traffic into folders and clearly not glancing at all of them.
Don't think I said before how much I appreciate you responding at all.
> On 13 August 2023 at 09:12, Chris Evans wrote:
> | I am putting this here as it may be of general and not just my own
> | interest.? I am currently running 4.3.1 on Ubuntu 22.04.2 LTS but am
> | hitting an issue that the magick package won't update because
> | libmagick++-dev won't update because, presumably, of two repositories
in
> | my sources.list disagreeing about versions required.? That's not so
>
> That can happen, and pinning can help. I would suggest to look at
'apt-cache
> polict nameofthepackagehere'. (See below for concrete example.)
So I try this.

root at Clevo2:/media/chris/Clevo_SSD2/Data/MyR/R/distill_blog/test2/_posts# 
apt-cache policy libmagick++-dev
libmagick++-dev:
 ? Installed: (none)
 ? Candidate: 8:6.9.11.60+dfsg-1.3ubuntu0.22.04.3+esm2
 ? Version table:
 ???? 8:6.9.11.60+dfsg-1.3ubuntu0.22.04.3+esm2 500
 ??????? 500 https://esm.ubuntu.com/apps/ubuntu jammy-apps-security/main 
amd64 Packages
 ??????? 500 https://esm.ubuntu.com/apps/ubuntu jammy-apps-security/main 
i386 Packages
 ???? 8:6.9.11.60+dfsg-1.3build2 500
 ??????? 500 http://mirror.infomaniak.ch/ubuntu jammy/universe amd64 
Packages
 ??????? 500 http://mirror.infomaniak.ch/ubuntu jammy/universe i386 Packages

And the key problems remains this (actually, it's got a bit worse):

root at Clevo2:/media/chris/Clevo_SSD2/Data/MyR/R/distill_blog/test2/_posts# 
apt-get install libmagick++-dev
Reading package lists... Done
Building dependency tree... Done
Reading state information... Done
Some packages could not be installed. This may mean that you have
requested an impossible situation or if you are using the unstable
distribution that some required packages have not yet been created
or been moved out of Incoming.
The following information may help to resolve the situation:

The following packages have unmet dependencies.
 ?libcairo2-dev : Depends: libfontconfig1-dev (>= 2.2.95)
 ???????????????? Depends: libfreetype6-dev (>= 2.1.10)
 ?libglib2.0-dev : Depends: libglib2.0-0 (= 2.72.4-0ubuntu1) but 
2.72.4-0ubuntu2.2 is to be installed
 ????????????????? Depends: libglib2.0-bin (= 2.72.4-0ubuntu1)
 ????????????????? Depends: libglib2.0-dev-bin (= 2.72.4-0ubuntu1)
 ?libmagickcore-6.q16-dev : Depends: libfreetype6-dev
 ?libwmf-dev : Depends: libfreetype-dev but it is not installable
E: Unable to correct problems, you have held broken packages.

Which brings me to Dirk's on the nail question:
> (Also: how come there are two libmagick++-dev, ie why do you have a 2nd
one,
> ie do you know which additional repo you turned on?)
Exactly!!! The weird thing is is that it was weeks back that I added a 
repo., or
I think I did, probably more than one about handling gpx files in 
Ubuntu/Debian.

But that was weeks before this started and as you see from that 
apt-cache policy
(if I understand it correctly), the repos involved seem to be the 
mainstream ones.
That's true when I do apt-cache policy on any of the packages that are 
being
complained about.

aptitude doesn't help and dpkg --get-selections | grep hold returns 
nothing (literally)
and none of the other things I could find on the web about these sorts 
of collisions
helped ... and I can't see that others have this problem (apart from 
some very old
reports for Ubuntu 14 and 16!)

Are there any more suggestions?? Are there good Debian/Ubuntu places 
where I should
be putting this?? If so, any advice about how to word it?!

Are there frightening radical steps I should take next?!

... and back to the human
>
> Cheers, Dirk
Cheers indeed.
>
> | Visiting Professor, UDLA, Quito, Ecuador & Honorary Professor,
> | University of Roehampton, London, UK.
>
> Nice. I got to workshop-keynote once in Guayaquil and regret not making
over
> to Quito.
Ah, I'm the reverse.? Ecuador seems to be sliding into a frightening 
narcogang
dominated state that previously it had always avoided.? I guess that puts my
problems in perspective.
>
-- 
Chris Evans (he/him)
Visiting Professor, UDLA, Quito, Ecuador & Honorary Professor, 
University of Roehampton, London, UK.
Work web site: https://www.psyctc.org/psyctc/
CORE site: http://www.coresystemtrust.org.uk/
Personal site: https://www.psyctc.org/pelerinage2016/