Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
  • Mar 29 18:58
    eliotmcintire synchronize #22
  • Mar 29 18:58

    eliotmcintire on installGitHubPackage

    minor bugfix tried to install g… bump version bugfix: rmDuplicatePkgs couldn'… and 3 more (compare)

  • Mar 18 18:24
    achubaty synchronize #24
  • Mar 18 18:24

    achubaty on misc-fixes

    bugfix: rmDuplicatePkgs couldn'… added globalVariables Merge pull request #25 from Cer… and 1 more (compare)

  • Mar 15 19:30

    eliotmcintire on development

    bugfix: rmDuplicatePkgs couldn'… added globalVariables Merge pull request #25 from Cer… (compare)

  • Mar 15 19:30
    eliotmcintire closed #25
  • Mar 15 19:13
    CeresBarros opened #25
  • Mar 09 05:29
    achubaty commented #24
  • Mar 09 05:28
    achubaty commented #24
  • Mar 09 05:28
    achubaty commented #24
  • Mar 09 05:25
    achubaty review_requested #24
  • Mar 09 05:25
    achubaty opened #24
  • Mar 09 05:23

    achubaty on misc-fixes

    respect env var 'R_USER_CACHE_D… (compare)

  • Mar 09 05:15

    achubaty on misc-fixes

    gitignore cleanup rm CRAN-RELEASE file on devel b… devel version bump; update NEWS… and 8 more (compare)

  • Mar 09 05:10
    achubaty opened #23
  • Mar 01 16:27

    eliotmcintire on development

    bump version (compare)

  • Feb 27 00:22

    eliotmcintire on development

    minor bugfix tried to install g… (compare)

  • Feb 26 17:39
    eliotmcintire synchronize #22
  • Feb 26 17:39

    eliotmcintire on installGitHubPackage

    minor when there were multiple … Merge branch 'development' into… (compare)

  • Feb 25 17:57

    eliotmcintire on development

    minor when there were multiple … (compare)

Alex Chubaty
@achubaty

it's putting things into packages_1, not packages as i would expect. and packages_2 is empty. i expect that the packages get installed to the location i specify (packages).

as soon as i try to run a sim, it tries reinstalling different versions of packages even though they should be correct, and it detaches and breaks the search path, breaking all subsequent package calls

completely unable to run anything
you will need to have that git repo on your machine of course, and that branch checked out
Alex Chubaty
@achubaty
Currently, several packages are loaded that conflict with installs. Detaching:
Rcpp, crayon, magrittr, sp, fpCompare, ps, assertthat, digest, gridBase, backports, pkgconfig, withr, R.methodsS3, gtable, DBI, fs, generics, colorspace, prettyunits, R6, rstudioapi, whisker, fastdigest, bit, fansi, RcppParallel, RApiSerialize, RColorBrewer, rprojroot, R.oo, rgdal, munsell, rgeos, ellipsis, bit64, memoise, raster, R.utils, purrr, lubridate, igraph, lifecycle, stringfish, processx, callr, qs, cli, vctrs, blob, desc, sessioninfo, tidyselect, scales, pillar, tibble, RSQLite, pkgbuild, pkgload, dplyr, reproducible, testthat, googledrive, usethis, ggplot2, quickPlot, devtools, SpaDES.core
it then proceeds to reinstall and reload these, which breaks and sometimes crashes the R session
i can't do this. i will install all 262 pkgs maunally (scripted at least)
Eliot McIntire
@eliotmcintire
What did you do?
Alex Chubaty
@achubaty
i tried using it based on tati and your examples. lol
it's really b0rked
Eliot McIntire
@eliotmcintire
I am trying it now. 2 obvious issues: Require is in the packageVersions.txt file ... and it is a version that conflicts. 2) devtools is being loaded. That has A LOT of dependencies. So, don't have devtools loaded. I am running stuff now. It will take a bit to catch up to you...
Of course, mine seems fine so far ... but I am just installing for the first time.
Alex Chubaty
@achubaty
this needs to be reproducible (i.e., i'm doing this for someone else) so they may or may not have packages installedi tried installing the version of Require that's in the package file with same result
Eliot McIntire
@eliotmcintire

@achubaty @tati-micheletti I pushed updates to Require ... running tests. Issues that I found:

  1. Require was not installed in the new libPath
  2. In my case, one package ("XML") couldn't be installed due to version of R conflict with package version. There is now a better error and suggestion about what to do.
  3. Suppressed some irrelevant warnings due to "locale"
  4. Removed base packages during installing from packageVersion.txt; this changes the double folder issue @achubaty saw
  5. The file ("standAloneSim.R") is not set up to be "reuseable"...:

... Do not load devtools -- it has lots of package dependencies and the versions will conflict with those versions in packageVersion.txt
... Do install Require into the new library (can do the original one too)

Tati Micheletti
@tati-micheletti
Thanks, @eliotmcintire! 1-4 understood, agreed and thankful for. point 5 (The file ("standAloneSim.R") is not set up to be "reuseable"...) I didn't understand...
Eliot McIntire
@eliotmcintire
Getting to passing on Github Actions... will merge into development shortly
Eliot McIntire
@eliotmcintire
Ok. I believe I have handled a few of the edge cases that were "found" in the above. Merged into development
Tati Micheletti
@tati-micheletti
Thanks, Eliot!! :)
Alex Chubaty
@achubaty
this is great, thanks Eliot. I'll give it another go and report back
Eliot McIntire
@eliotmcintire
@achubaty I will try to submit Require to CRAN today or tomorrow (I suspect before reproducible because of dependency chain...)
Eliot McIntire
@eliotmcintire
Dear maintainer,
thanks, package Require_0.0.9.tar.gz is on its way to CRAN.
Best regards,
CRAN teams' auto-check service
Eliot McIntire
@eliotmcintire
Wow. It is up already. That was fast! https://cran.r-project.org/web/packages/Require/index.html
Alex Chubaty
@achubaty
i tried standALoneSim.R again and it still fails. Josie is also having issues (though hers are different from mine)
(tried using the latest version of Require)
raster is still being re-installed to a different version than the one loaded then this new version is being forced to be loaded, which cause the same breakage i was seeing before
Eliot McIntire
@eliotmcintire
Hard to reproduce... presumably you are starting from a clean session with no packages loaded?
Eliot McIntire
@eliotmcintire
@achubaty Did you comment out the loading of devtools in the script?
Alex Chubaty
@achubaty

not sure what you mean. i'm not loading devtools.

also, i see you reverted my change that uses the development branch of Require (instead of tmp1 branch). was that intentional?

Eliot McIntire
@eliotmcintire
  1. By installing development branch, with devtools, you are loading devtools if you run the script as is. If you do manual intervention, then it might be fine.
  2. Require 0.0.9 is on CRAN as of now (I have to quickly send an update for 0.0.10 due to error on one of the flavours), so don't need dev version at all.
Eliot McIntire
@eliotmcintire
I just pushed again... not sure what the devtools install was still there for.

@achubaty I ran the Require::Require line 2x and it worked after the 2nd time.

* DONE (usefulFuns)
All packages appear to have installed correctly

Because of transient failures with network... there used to be a message saying something to the effect of "run it again"... that seems gone now.

For the record, this worked and installed everything:
options(repos = c(CRAN = "https://cloud.r-project.org"),  
        Require.RPackageCache = "~/binaryRPackages")      ## change this to your tastes

if (!requireNamespace("Require", quietly = TRUE))
  install.packages("Require") # use version 0.0.9

# 0. Install working packages
library("Require")

pkgPath <- "./packages"
setLibPaths(pkgPath)

Require::Require(packageVersionFile = "./packageVersions.txt", standAlone = TRUE) # failed on some 1st tiem
Require::Require(packageVersionFile = "./packageVersions.txt", standAlone = TRUE)
Alex Chubaty
@achubaty
thanks Eliot, i'll give it another go
Eliot McIntire
@eliotmcintire
I deleted all packages, and reran (using the binary cache) and it installed everything correctly the first try (in <5 minutes).
Alex Chubaty
@achubaty
From Josie: And Require tells me package path doesn't exist even though it does.
Require::Require(packageVersionFile = "./packageVersions.txt", standAlone = TRUE)
Error in checkPath(path, create = FALSE) :
  Specified path C:/Users/HughesJo/Documents/gitprojects/NWT/packages doesn't exist
Eliot McIntire
@eliotmcintire
Just found that too... pushing a fix ... and it will be on CRAN within 24 hours.
The fix is in Require@development now...
Alex Chubaty
@achubaty
thanks
:)
Eliot McIntire
@eliotmcintire
I believe there are some weird edge cases where there was a version of a package that was installable for a previous version of R (remember, we may not be using the same version of R), but is not installable for the user's current version of R. The messaging at the end should offer a solution ... namely, just install the package with install.packages since in most cases the latest package is compatible with a previous version.
Also, just got a "Unix- only package" error on a Windows install.
I think both of these cases are acceptable for now.
Eliot McIntire
@eliotmcintire
Require 0.0.10 is now on CRAN. It has this error fixed:
Require::Require(packageVersionFile = "./packageVersions.txt", standAlone = TRUE)
Error in checkPath(path, create = FALSE) :
  Specified path C:/Users/HughesJo/Documents/gitprojects/NWT/packages doesn't exist
1 reply
Tati Micheletti
@tati-micheletti
Yay!! :D
Eliot McIntire
@eliotmcintire
Require seems to work fine with the RStudio Binary Package Manager (e.g., options(repos = list(CRAN = "https://packagemanager.rstudio.com/all/__linux__/focal/latest")). It correctly installs binary packages.
Alex Chubaty
@achubaty
awesome!! for current packages only?
Eliot McIntire
@eliotmcintire
Still untested is when Require is using a PackageSnapshop.txt file, meaning it needed to install specific versions of packages. Reports suggest that is failing. This is not surprising because it seems to have an idiosyncratic folder structure.
Alex Chubaty
@achubaty
@eliotmcintire not sure why these won't combine properly in covr but here are the individual coverage reports:
> cov1 <- covr::package_coverage(type = "tests")
> cov2 <- covr::package_coverage(type = "examples")
> cov3 <- covr::package_coverage(type = "vignettes")
> cov1
Require Coverage: 0.20%
R/extract.R: 0.00%
R/helpers.R: 0.00%
R/pkgDep.R: 0.00%
R/pkgDepAlt.R: 0.00%
R/pkgSnapshot.R: 0.00%
R/Require-helpers.R: 0.00%
R/Require.R: 0.00%
R/RequireOptions.R: 0.00%
R/setLibPaths.R: 0.00%
R/CRAN.R: 5.26%
R/zzz.R: 5.56%
> cov2
Require Coverage: 22.62%
R/Require.R: 0.00%
R/RequireOptions.R: 0.00%
R/pkgDepAlt.R: 4.61%
R/zzz.R: 5.56%
R/Require-helpers.R: 8.01%
R/setLibPaths.R: 34.48%
R/pkgSnapshot.R: 37.50%
R/helpers.R: 59.70%
R/pkgDep.R: 61.45%
R/CRAN.R: 66.67%
R/extract.R: 89.19%
> cov3
Require Coverage: 0.09%
R/extract.R: 0.00%
R/helpers.R: 0.00%
R/pkgDep.R: 0.00%
R/pkgDepAlt.R: 0.00%
R/pkgSnapshot.R: 0.00%
R/Require-helpers.R: 0.00%
R/Require.R: 0.00%
R/RequireOptions.R: 0.00%
R/setLibPaths.R: 0.00%
R/CRAN.R: 4.17%
R/zzz.R: 5.56%
examples cower most of the code, whereas tests apparently don't???
i don't believe it
Eliot McIntire
@eliotmcintire

Unfortunately, we see:

New things in home:
._RPackageCache
._RPackageCache/4.1
._RPackageCache/4.1/data.table_1.13.6_R_x86_64-pc-linux-gnu.tar.gz
._RPackageCache/4.1/data.table_1.13.6.tar.gz
._RPackageCache/4.1/remotes_2.2.0_R_x86_64-pc-linux-gnu.tar.gz
._RPackageCache/4.1/remotes_2.2.0.tar.gz
._RPackageCache/4.1/Require_0.0.10_R_x86_64-pc-linux-gnu.tar.gz
._RPackageCache/4.1/Require_0.0.10.tar.gz

Well, clearly you caused it in :

/data/rsync/PKGS/Require$ grep -r _RPackageCache .
./R/RequireOptions.R: Require.RPackageCache = NULL,

"~/._RPackageCache", # nolint

./tests/testit/test-packagesLong.R:
Require.RPackageCache = "~/._RPackageCache",
./tests/testit/test-pkgSnapshot.R: "Require.RPackageCache" =
"~/._RPackageCache/",
./tests/testit/test-1packages.R:
Require.RPackageCache = "~/._RPackageCache",

And storing files there is not permitted.

Please re-read the CRAN policies that also include a suggestion:

For R version 4.0 or later (hence a version dependency is required or
only conditional use is possible), packages may store user-specific
data, configuration and cache files in their respective user
directories obtained from tools::R_user_dir(), provided that by
default sizes are kept as small as possible and the contents are
actively managed (including removing outdated material).

Please fix and resubmit.