conda env create -f bin/INSPIIRED.conda.yml --> ResolvePackageNotFound

[jwhite007]

I would really appreciate some help with the following. Thanks.

System

• Red Hat Enterprise Linux Server release 6.6 (Santiago)
• conda 4.9.2 (miniconda)
  • channels: r, conda-forge, bioconda, defaults

Command

$ conda env create -f bin/INSPIIRED.conda.yml
Collecting package metadata (repodata.json): done
Solving environment: failed

ResolvePackageNotFound:
  \- mkl==11.3.3=0
  \- pango==1.39.0=0
  \- glib==2.43.0=2
  \- numpy==1.11.0=py27_1
  \- freetype==2.5.5=1
  \- git==2.6.4=0
  \- harfbuzz==0.9.39=0
  \- r-getopt==1.20.0=r3.2.2_0
  \- cffi==1.6.0=py27_0
  \- tk==8.5.18=0
  \- libpng==1.6.17=0
  \- jpeg==8d=0
  \- pyopenssl==0.16.0=py27_0
  \- curl==7.49.0=0
  \- biopython==1.67=np111py27_0
  \- cairo==1.12.18=6
  \- libtiff==4.0.6=2
  \- cryptography==1.4=py27_0
  \- pycparser==2.14=py27_1
  \- libxml2==2.9.3=0
  \- python==2.7.11=0
  \- pcre==8.31=0
  \- readline==6.2=2

I then do the following which results in a much shorter list:

$ export CONDA_RESTORE_FREE_CHANNEL=1
$ conda env create -f bin/INSPIIRED.conda.yml
Collecting package metadata (repodata.json): done
Solving environment: failed

ResolvePackageNotFound:
  - libxml2==2.9.3=0
  - glib==2.43.0=2
  - r-getopt==1.20.0=r3.2.2_0

I then update bin/INSPIIRED.conda.yml with versions that are found in my channels:
- libxml2=2.9.3=6
- glib=2.43.0=1
- r-getopt=1.20.0

which results in what seems like hundreds of Package conflicts, e.g.:

Package r-boot conflicts for:
r==3.2.2=0 -> r-recommended==3.2.2 -> r-boot
r-boot==1.3_17=r3.2.2_0a
r-recommended==3.2.2=r3.2.2_0 -> r-boot

[gmilton-425]

I am having the same issue and would like help with the ResolvePackageNotFound.

[ressy]

My first thought was missing channels from the condarc file, but as you mentioned, @jwhite007, you already have the necessary ones included, and I see the same problem when I try it just now.

I think the main issue may just be that these pinned versions are too old and have been dropped by whatever channel is supplying them. For example I see the earliest pango available in the defaults is 1.40.11 instead of the 1.39.0 listed. Some of these can be found by searching conda-forge as well as defaults and bioconda, but not quite all. If I raise the versions just enough to match available packages, a large number of other pinned packages then show conflicts with these higher versions of their dependencies.

I think what this really requires is a careful update to the list of truly-needed-by-INSPIIRED packages, since that INSPIIRED.conda.yml file with exact versions and build IDs is really just a snapshot of what existed in a particular known-working install, and the conda channels have since left many of those packages behind. But I don't think that sort of minimal environment file is included in this repository. @helixscript @cnobles any thoughts? (I'm just pitching in my 2 cents from conda experience here, but you'd know better than I would, as far as INSPIIRED goes.)

[jwhite007]

@ressy

Thank you so much for your attention to this. You rock! I've got some more information with which I will update my original post above. Cheers!

[helixscript]

Hi jwhite and co,

The required CONDA environment continues to be a weakness for this software.
One common problem is that the software uses a number of dplyr functions which have
changed over the years since dplyr 0.5 and both CONDA and R have a tendency to update
dplyr and other packages behind the scenes. I believe the best path forward would be to
reduce our reliance on CONDA and offer an archive of specific R source packages and an
install script while only relying on CONDA for R and compilers. This will take a bit of time
to complete and hope to update the group by the end of the week.

Best - John

[jwhite007]

@helixscript, thanks for the reply and also for looking into this. It's much appreciated.

[gmilton-425]

Thank you for looking into this, is there any update?

[jwhite007]

@helixscript, If there's anything at all that I can do to help, please let me know. I can start by removing R and all of the R packages from the yml, and then installing R and the packages separately. Would this help? Thanks.

[gmilton-425]

Hello, @helixscript is there any update on this issue? You had previously said you may be able to provide a collection of Linux
R source packages archived for this project which we will make available as a potential path forward for satisfying the required environment.

[Gmilton425]

@helixscript @jwhite007 @chasberry Has the been any movement on this? @cnobles

[lyoussar]

Any update? I am running into the same issue.