Closed chlebowa closed 1 year ago
Thanks, does this happen locally as well?
I am unable to run the app locally with restore_and_run
as helios
Depends
on R >= 4.3
and R 4.3 was not published for Ubuntu.
Running on my own installation (helios
dependency lifted) I get this error:
Warning: Error in [[<-: 5 elements in value to replace 2 elements
3: runApp
2: print.shiny.appobj
1: <Anonymous>
Warning: Error in [[<-: 5 elements in value to replace 2 elements
3: runApp
2: print.shiny.appobj
1: <Anonymous>
Warning: Error in [[<-: 5 elements in value to replace 2 elements
126: <Anonymous>
125: stop
124: pca_result
123: <reactive>
107: show_matrix_pca
106: exprFunc
105: widgetFunc
104: ::
htmlwidgets
shinyRenderWidget
103: func
90: renderFunc
89: renderFunc
85: renderFunc
84: output$teal-main_ui-root-pca_plot-module-table_pca
3: runApp
2: print.shiny.appobj
1: <Anonymous>
Warning: Error in [[<-: 5 elements in value to replace 2 elements
225: <Anonymous>
224: stop
223: pca_result
222: <reactive>
220: .func
217: contextFunc
216: env$runWith
209: ctx$run
208: self$.updateValue
206: plot_r
204: <reactive>
202: .func
199: contextFunc
198: env$runWith
191: ctx$run
190: self$.updateValue
188: plot_type
187: <reactive>
185: .func
182: contextFunc
181: env$runWith
174: ctx$run
173: self$.updateValue
171: plot_reactive
170: renderPlot
168: func
128: drawPlot
114: <reactive:plotObj>
98: drawReactive
85: renderFunc
84: output$teal-main_ui-root-pca_plot-module-plot_pca-plot_main
3: runApp
2: print.shiny.appobj
1: <Anonymous>
Wait this is not depending on helios, just hermes. Staged dependencies should help with installation
My mistake. The error I got here is of the same nature but comes from S4Arrays
:
ERROR: this R is version 4.2.2, package 'S4Arrays' requires R >= 4.3.0
Error: install of package 'S4Arrays' failed [error code 1]
In addition: Warning message:
This project is configured to use Bioconductor 3.17, which is not compatible with R 4.2.2.
Use 'renv::init(bioconductor = "3.17")' to re-initialize this project with the appropriate Bioconductor release.
restore_and_run
uses renv
, I don't think sd will be of much help.
Hm interesting. Does this mean we also depend now on R 4.3 with this package?
I'm actually not sure. The the renv
in teal.gallery
wants Bioconductor 3.17
which contains DelayedArray 0.26.6
which requires S4Arrays
, which requires R 4.3
. I have Bioconductor 3.16 with
DelayedArray`0.24.0' and no
S4Arrays` and the application runs.
Ah ok. Could that be a mistake on the renv config then? Because teal.gallery would be supposed to support the current release on 4.2.2 or not?
@vedhav can you chip in?
I see the problem. As you pointed out, I tried to restore in R 4.2.2, and it failed because of the S4Arrays
installation.
However, I am able to install hermes
from R 4.2.2 (without renv.lock) because it installs an older version from BioC which does not have the S4Arrays
dependency (As it installs the SummarizedExperiment version 1.28.0). The problem arises because the renv.lock was generated with a newer version of the hermes
package with an updated SummarizedExperiment
version. So we could restore the renv.lock file with an older R version (4.2) with BioC 3.16
to fix this or just roll back the SummarizedExperiment
version from GitHub.
But, I want to point out that maintaining an older version of renv.lock
files are tricky. Currently, the plan is to make sure that we update the renv.lock file with the latest packages and make sure that the app works as expected.
I would think that there is a general inconsistency here between the deployment environment R version (4.2.2) and the renv lock file (needs BioC 3.17 which needs R 4.3). Or not?
The deployment environment R version is 4.3.1 (2023-06-16) which is in line with the renv.lock file.
But on Ocean we don't have 4.3 yet or have we?
Not sure about Ocean. But, the deployment is done by the CI image ghcr.io/insightsengineering/rstudio_4.3.1_bioc_3.17:latest
here's the workflow that does that.
I confirm. I've followed steps @chlebowa described and app fails when range is changed. I'm on R 4.3.0
I think I've found the source of the problem. Looks like it is on us (coredev), I will continue tomorrow.
In RNA-seq app:
app crashes