reshape2: cast/melt require en_US.UTF-8, other assorted issues.

[hexylena]

locale

en_US isn't my locale, isn't installed at a system level, and isn't installed as part of the conda env.

cc @bgruening looks like you were the last to touch these.

Variable not found

## Setup R error handling to go to stderr                                                         
options(show.error.messages=F, error=function(){cat(geterrmessage(),file=stderr());q("no",1,F)})  
loc <- Sys.setlocale("LC_MESSAGES", "en_US.UTF-8")                                                
                                                                                                  
## Import library                                                                                 
library("reshape2")                                                                               
                                                                                                  
input <- read.csv('$input', sep='\t', header=TRUE)                                                
cinput <- dcast(input, ... ~ variable)                                                            
write.table(cinput, "output.tabular", sep="\t", quote=FALSE, row.names=FALSE)                     

I'm not sure where variable should've come from. I suppose it should've been annotated in my dataset but then that assumption needs to be made explicit and communicated to the user.

Dependency Issues

I also had to manually update R in that conda env to even get the tools to work, I don't understand how that happened though. The original resolution of r-reshape2 at version 1.4.2 included an r-base that depended on libgfortran.so.3 which isn't installed alongside that version of r? deeply unhelpful. After manually conda install 'r-base>3.6' can the tool work at all, due to the lack of libgfortran3 (only 5 was installed). If we can pin r-base>3.6 it could be helpful maybe for folks that want to install this.

Help Section

The help section of this tool would also greatly benefit from two quick tables showing the before/after of the operation. here are some sample tables for whoever wants to update this tool.

station ozone      wind     temp
1 23.61538 11.622581 65.54839
2 29.44444 10.266667 79.10000

and

station variable value
1 ozone 23.61538 
1 wind 11.622581
1 temp 65.54839
2 ozone 29.44444
2 wind 10.266667
2 temp 79.10000

[hexylena]

Additionally I have a dataset with missing values. dcast has a terrible worst case here, it'll use a different aggregation function, which is rather unhelpful.

  amr    variable value
  <chr>  <chr>    <dbl>
1 tet(M) WF1      100  
2 tet(M) WF3       96  
3 tet(M) WF3       97.4

will produce

> d %>% filter(amr=="tet(M)") %>% dcast(... ~ variable)
Aggregation function missing: defaulting to length
     amr WF1 WF3
1 tet(M)   1   2

it would be helpful to be able to set fun.aggregate to things like first or mean or sum.