search

flr / FLXSA

The FLXSA package
http://flr-project.org/FLXSA
2 stars 3 forks source link

XSA exception in Rstudio 1.2.xx version #6

Closed zenn1989 closed 3 years ago

zenn1989 commented 5 years ago

Hello. If you try to perform typical XSA in new version FLXSA package with Rstudio 1.2.xx then you will got error:

Error in lapply(assess@index, FLQuant) : 
  negative length vectors are not allowed

You can simply reproduce this bug with test data (ple4): 

data(ple4)
data(ple4.indices)
test <- FLXSA(ple4, ple4.indices)

Is there any solution exist or i should rollback to previous version? Problem occured only in Windows OS and Rstudio 1.2.xx or newst.

UPD: another members in comments found problem source in flxsa.R in fqs() function in lapply() calling from res <-fqs(.Call("runFLXSA", stock, indices, control, diag.flag))

zenn1989 commented 5 years ago

Update: the first run of XSA() going ok, but second and then coming - fails. Screenshot_3

zenn1989 commented 5 years ago

Sorry, my bad. It look's like a Rstudio 1.2.xx bug. Run XSA in RGUI - all is ok.

iagomosqueira commented 5 years ago

Did you identify what is causing this problem in RStudio? Would be useful to know and see if it can be avoided.

zenn1989 commented 5 years ago

Did you identify what is causing this problem in RStudio? Would be useful to know and see if it can be avoided.

No, i have no idea why it happens, but in Rstudio 1.2.xx always return error after 2nd and coming calls FLXSA. In Rstudio 1.1.xx all works (also in RGUI) ok.

hgerritsen commented 4 years ago

I don't think it's (just) an RStudio problem. I get the same issue when i use R 3.6.1 in RStudio 1.2.5 but it works fine with R 3.5.1 with the same version of RSudio

iagomosqueira commented 4 years ago

But this still only happens in Rstudio, right? What do you see if you call for the definition of lapply?

lapply
showMethods("lapply")
hgerritsen commented 4 years ago

Yes, it works fine with r 3.6.1 outside of RStudio, but somehow it does work in Rstudio If you use an older version of R.

showMethods("lapply") Function: lapply (package base) X="ANY" X="FLComps" (inherited from: X="FLlst") X="FLIndices" (inherited from: X="FLlst") X="FLlst" X="FLQuants" (inherited from: X="FLlst") X="list" (inherited from: X="ANY")

From: Iago Mosqueira [mailto:notifications@github.com] Sent: Tuesday 11 February 2020 20:56 To: flr/FLXSA FLXSA@noreply.github.com Cc: Hans Gerritsen hans.Gerritsen@Marine.ie; Comment comment@noreply.github.com Subject: Re: [flr/FLXSA] XSA broken in 2.6.2 release (#6)

But this still only happens in Rstudio, right? What do you see if you call for the definition of lapply?

lapply

showMethods("lapply")

— You are receiving this because you commented. Reply to this email directly, view it on GitHubhttps://github.com/flr/FLXSA/issues/6?email_source=notifications&email_token=AHA5GPOBLT4WORMIQZVUXZDRCMGEXA5CNFSM4IOCPJT2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOELOA4UA#issuecomment-584846928, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AHA5GPKAATURA53WKKJPTULRCMGEXANCNFSM4IOCPJTQ.

iagomosqueira commented 4 years ago

I am afraid I am completely lost here. Not sure it will tell much, but could you call

traceback()

after the error and report the result?

hgerritsen commented 4 years ago

library(FLXSA)

data(ple4) data(ple4.indices) res <- FLXSA(ple4, ple4.indices) res <- FLXSA(ple4, ple4.indices) Error in lapply(assess@index, FLQuant) : negative length vectors are not allowed traceback() 8: lapply(assess@index, FLQuant) 7: initialize(value, ...) 6: initialize(value, ...) 5: new("FLQuants", lapply(assess@index, FLQuant)) 4: fqs(.Call("runFLXSA", stock, indices, control, diag.flag)) 3: .local(stock, indices, ...) 2: FLXSA(ple4, ple4.indices) 1: FLXSA(ple4, ple4.indices)

note that if i then try: lapply(res@index,FLQuant) this works fine Also, i tried 32 bit R in RStudio and that also works fine.

hgerritsen commented 4 years ago

If it's any help: the error does not occur if i do this: res <- FLXSA(ple4, ple4.indices,diag.flag=F)

But when i stepped through the FLXSA function, both of these cause the 'negative length vector' error .Call("runFLXSA", stock, indices, control, diag.flag=T) Error: negative length vectors are not allowed .Call("runFLXSA", stock, indices, control, diag.flag=F) Error: negative length vectors are not allowed

colinpmillar commented 4 years ago

Also getting this error running in the ICES TAF server. Currently using quick fix from @hgerritsen. But this means we do not get any diagnostics. Will send our traceback also.

colinpmillar commented 4 years ago

Hi, the error is coming from the C code:

diag.flag <- TRUE
res <- .Call("runFLXSA", stock, indices, control, diag.flag)     
Error: negative length vectors are not allowed

will have a quick look at the C code, but I have not debuged C++ code use with R before...

colinpmillar commented 4 years ago

so it errors when you pass an FLIndices object, but not with an FLIndex

 res <- .Call("runFLXSA", xsa.stock, xsa.indices[1], xsa.control, diag.flag)  
Error: negative length vectors are not allowed
res <- .Call("runFLXSA", xsa.stock, xsa.indices[[1]], xsa.control, diag.flag)
colinpmillar commented 4 years ago

sorry - last post was not correct:

res <- .Call("runFLXSA", xsa.stock, xsa.indices[[1]], xsa.control, d 
    iag.flag)
res
[1] "Error in FLIndices"

passing an FLIndex returns an error also, but an expected one.

I have tested on previous versions of FLXSA, but not previous versions of FLCore, or R. But older versions (2 years ago) give same error.

iagomosqueira commented 4 years ago

I am testing the version currently in github using

install_github("flr/FLash")

library(FLXSA)

data(ple4)
data(ple4.indices)

res <- FLXSA(ple4, ple4.indices)

and I get no error. If anybody can replicate this I will release a new version.

This is on R3.6.2 and Linux.

iagomosqueira commented 4 years ago

Also works on R 3.6.1 in Windows 10

colinpmillar commented 4 years ago

Will give it a go!

iagomosqueira commented 4 years ago

I have tried with the two values of diag.flag

res <- FLXSA(ple4, ple4.indices,diag.flag=F)
res <- FLXSA(ple4, ple4.indices,diag.flag=T)
colinpmillar commented 4 years ago

Yep that worked for me too - but it still fails for the objects attached... don't want to go down the wormhole of corupt objects... but have uploaded in case you have time to check with them

xsa.zip

download.file("https://github.com/flr/FLXSA/files/4295135/xsa.zip", destfile = "xsa.zip")
# trying URL 'https://github.com/flr/FLXSA/files/4295135/xsa.zip'
# Content type 'application/x-zip-compressed' length 7713 bytes
# downloaded 7713 bytes

unzip("xsa.zip")
library(FLXSA)
# Loading required package: FLCore
# Loading required package: lattice
# Loading required package: iterators
# FLCore (Version 2.6.14, packaged: 2019-11-18 21:54:10 UTC)
# Loading required package: FLAssess

load("xsa.rData")
xsa <- FLXSA(xsa.stock, xsa.indices, xsa.control, diag = TRUE)
# Error in lapply(assess@index, FLQuant) : 
#   negative length vectors are not allowed
# sessioninfo::session_info()
# - Session info ----------------------------------------------------------
#  setting  value
#  version  R version 3.6.3 (2020-02-29)
#  os       Windows 10 x64
#  system   x86_64, mingw32
#  ui       RTerm
#  language (EN)
#  collate  English_United Kingdom.1252
#  ctype    English_United Kingdom.1252
#  tz       Europe/Paris
#  date     2020-03-05

# - Packages ------------------------------------------------------------------- 
# ! package     * version  date       lib source
#    assertthat    0.2.1    2019-03-21 [1] CRAN (R 3.6.0)
#    cli           2.0.2    2020-02-28 [1] CRAN (R 3.6.2)
#    crayon        1.3.4    2017-09-16 [1] CRAN (R 3.6.0)
#    fansi         0.4.1    2020-01-08 [1] CRAN (R 3.6.2)
#    FLAssess    * 2.6.3    2019-12-11 [1] local
#    FLCore      * 2.6.14   2019-11-18 [1] flr (R 3.6.1)
#    FLXSA       * 2.6.3    2019-11-18 [1] local
#    glue          1.3.1    2019-03-12 [1] CRAN (R 3.6.0)
#    iterators   * 1.0.12   2019-07-26 [1] CRAN (R 3.6.1)
#    lattice     * 0.20-40  2020-02-19 [1] CRAN (R 3.6.3)
#    MASS          7.3-51.5 2019-12-20 [1] CRAN (R 3.6.2)
#  R Matrix        1.2-18   <NA>       [2] <NA>
#    sessioninfo   1.1.1    2018-11-05 [1] CRAN (R 3.6.2)
#    withr         2.1.2    2018-03-15 [1] CRAN (R 3.6.0)
colinpmillar commented 4 years ago

Ha - good luck finding this bug!! error me once, and you wont error me a second time

# R version 3.6.3 (2020-02-29) -- "Holding the Windsock"
# Platform: x86_64-w64-mingw32 (64-bit)

library(FLXSA)
# Loading required package: FLCore
# Loading required package: lattice
# Loading required package: iterators
# FLCore (Version 2.6.14, packaged: 2019-11-18 21:54:10 UTC)
# Loading required package: FLAssess

load("xsa.rData")
xsa <- FLXSA(xsa.stock, xsa.indices, xsa.control, diag = TRUE)
# Error in lapply(assess@index, FLQuant) : 
#   negative length vectors are not allowed

xsa <- FLXSA(xsa.stock, xsa.indices, xsa.control, diag = TRUE)
# NO ERROR AND RUNS FINE!

you can even reload the data and it will still work. I tried running it up to 7 times and it seems to fail randomly.

colinpmillar commented 4 years ago

So I get it to work using just R. But when I run using radian (python clone of R) and Rstudio it fails.

So another workaround is to do:

res <- try(FLXSA(ple4, ple4.indices, diag = TRUE))
while(inherits(res, "try-error")) {
  res <- try(FLXSA(ple4, ple4.indices, diag = TRUE))  
}

this atleast gets your diagnostics

drfinlayscott commented 4 years ago

Random fails worry me. Smells like a C++ memory allocation thing which is a potential time bomb.

Do we know if it's failing before it reaches the .Call() to the C++ code, or during it?

Has the C++ code ever been updated since Kell first wrote it? I'm not offering, by the way.

Cheers from Auckland!

On Fri, 6 Mar 2020, 13:06 Colin Millar, notifications@github.com wrote:

Ha - good luck finding this bug!! error me once, and you wont error me a second time

R version 3.6.3 (2020-02-29) -- "Holding the Windsock"# Platform: x86_64-w64-mingw32 (64-bit)

library(FLXSA)# Loading required package: FLCore# Loading required package: lattice# Loading required package: iterators# FLCore (Version 2.6.14, packaged: 2019-11-18 21:54:10 UTC)# Loading required package: FLAssess

load("xsa.rData")xsa <- FLXSA(xsa.stock, xsa.indices, xsa.control, diag = TRUE)# Error in lapply(assess@index, FLQuant) : # negative length vectors are not allowed xsa <- FLXSA(xsa.stock, xsa.indices, xsa.control, diag = TRUE)# NO ERROR AND RUNS FINE!

you can even reload the data and it will still work. I tried running it up to 7 times and it seems to fail randomly.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/flr/FLXSA/issues/6?email_source=notifications&email_token=AAKB7FRWHKX6FJLQNF7Z3KLRGA46PA5CNFSM4IOCPJT2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEN7BHUQ#issuecomment-595465170, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAKB7FTZSECRW7CEUSX6RFLRGA46PANCNFSM4IOCPJTQ .

drfinlayscott commented 4 years ago

Looking at Colin's error, it's an R error rather than a C++ crash. Does this occur before the call to C++ or is it when it returns from C++ to R, i.e. some problem with creating the Sexp objects and sending them to R?

On Fri, 6 Mar 2020, 13:23 Finlay Scott, drfinlayscott@gmail.com wrote:

Random fails worry me. Smells like a C++ memory allocation thing which is a potential time bomb.

Do we know if it's failing before it reaches the .Call() to the C++ code, or during it?

Has the C++ code ever been updated since Kell first wrote it? I'm not offering, by the way.

Cheers from Auckland!

On Fri, 6 Mar 2020, 13:06 Colin Millar, notifications@github.com wrote:

Ha - good luck finding this bug!! error me once, and you wont error me a second time

R version 3.6.3 (2020-02-29) -- "Holding the Windsock"# Platform: x86_64-w64-mingw32 (64-bit)

library(FLXSA)# Loading required package: FLCore# Loading required package: lattice# Loading required package: iterators# FLCore (Version 2.6.14, packaged: 2019-11-18 21:54:10 UTC)# Loading required package: FLAssess

load("xsa.rData")xsa <- FLXSA(xsa.stock, xsa.indices, xsa.control, diag = TRUE)# Error in lapply(assess@index, FLQuant) : # negative length vectors are not allowed xsa <- FLXSA(xsa.stock, xsa.indices, xsa.control, diag = TRUE)# NO ERROR AND RUNS FINE!

you can even reload the data and it will still work. I tried running it up to 7 times and it seems to fail randomly.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/flr/FLXSA/issues/6?email_source=notifications&email_token=AAKB7FRWHKX6FJLQNF7Z3KLRGA46PA5CNFSM4IOCPJT2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEN7BHUQ#issuecomment-595465170, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAKB7FTZSECRW7CEUSX6RFLRGA46PANCNFSM4IOCPJTQ .

colinpmillar commented 4 years ago

Have tried a small test script: https://gist.github.com/colinpmillar/2bc67877dcea0c33e7a087f5aa1e6236

So far finding that if you run .Call("runFLXSA"... it crashes out every time in rstudio and radian. However, if you run FLXSA(... it fails more often at the start and seems to get less likely as you go on... though I have not tested this. I was going to set up a bash script to try and explore.

In any case, now that I can relaiably create the error message via:

library(FLXSA)
data(ple4)
data(ple4.indices)
.Call("runFLXSA", ple4, ple4.indices[1], FLXSA.control(), TRUE)

it might help route out the cause

iagomosqueira commented 4 years ago

I will have another go. I am also fearful of memory allocation being behind it.

laurieKell commented 4 years ago

It could be a C++ issue. The diags flag turns off all the diagnostic output when running simulations. There are many arrays in C++ that have to be returned otherwise.

see line 252 onwards in flxsa.R.

Maybe dump assess@index to a file. Then when R crashes have a look at the object.

i.e.

 fqs <- function(assess) {

      save(a,file="a.RData")

      assess@index <- new("FLQuants", lapply(assess@index,FLQuant))       assess@index.hat <- new("FLQuants", lapply(assess@index.hat,FLQuant))       assess@index.var <- new("FLQuants", lapply(assess@index.var,FLQuant))       assess@index.res <- new("FLQuants", lapply(assess@index.res,FLQuant))       assess@q.hat <- new("FLQuants", lapply(assess@q.hat,FLQuant))       assess@q2.hat <- new("FLQuants", lapply(assess@q2.hat,FLQuant))

      if (validObject(assess))         return(assess)       else         stop("not valid")     }

Laurie

On 06/03/2020 00:27, Finlay Scott wrote:

Looking at Colin's error, it's an R error rather than a C++ crash. Does this occur before the call to C++ or is it when it returns from C++ to R, i.e. some problem with creating the Sexp objects and sending them to R?

On Fri, 6 Mar 2020, 13:23 Finlay Scott, drfinlayscott@gmail.com wrote:

Random fails worry me. Smells like a C++ memory allocation thing which is a potential time bomb.

Do we know if it's failing before it reaches the .Call() to the C++ code, or during it?

Has the C++ code ever been updated since Kell first wrote it? I'm not offering, by the way.

Cheers from Auckland!

On Fri, 6 Mar 2020, 13:06 Colin Millar, notifications@github.com wrote:

Ha - good luck finding this bug!! error me once, and you wont error me a second time

R version 3.6.3 (2020-02-29) -- "Holding the Windsock"# Platform:

x86_64-w64-mingw32 (64-bit)

library(FLXSA)# Loading required package: FLCore# Loading required package: lattice# Loading required package: iterators# FLCore (Version 2.6.14, packaged: 2019-11-18 21:54:10 UTC)# Loading required package: FLAssess

load("xsa.rData")xsa <- FLXSA(xsa.stock, xsa.indices, xsa.control, diag = TRUE)# Error in lapply(assess@index, FLQuant) : # negative length vectors are not allowed xsa <- FLXSA(xsa.stock, xsa.indices, xsa.control, diag = TRUE)# NO ERROR AND RUNS FINE!

you can even reload the data and it will still work. I tried running it up to 7 times and it seems to fail randomly.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub

https://github.com/flr/FLXSA/issues/6?email_source=notifications&email_token=AAKB7FRWHKX6FJLQNF7Z3KLRGA46PA5CNFSM4IOCPJT2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEN7BHUQ#issuecomment-595465170, or unsubscribe

https://github.com/notifications/unsubscribe-auth/AAKB7FTZSECRW7CEUSX6RFLRGA46PANCNFSM4IOCPJTQ .

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/flr/FLXSA/issues/6?email_source=notifications&email_token=AAM2VJ6LDJ5MBJR5FBOLTZDRGA7O7A5CNFSM4IOCPJT2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEN7NDGI#issuecomment-595513753, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAM2VJ3MRILJM4QPVF6TKB3RGA7O7ANCNFSM4IOCPJTQ.

laurieKell commented 4 years ago

looks like there is something wrong with object@index[[i]]

What is the value of i and what does object@index[[i]] look like

On 06/03/2020 08:16, Colin Millar wrote:

Something that might be interesting. I don't get the error message when running in R, but when you inspect the element, you get:

res <- .Call("runFLXSA",ple4,ple4.indices[1], FLXSA.control(),TRUE)

res An object of class "FLXSA"

Name: Description: Index.name : Call : new("FLAssess") Quant: age Dims: age year unit season area iter 10 62 1 1 1 1

Range: min max minyear maxyear 0 0 1 1

survivors : [10 62 1 1 1 1 ],units = NA se.int : [9 1 1 1 1 1 ],units = NA se.ext : [9 1 1 1 1 1 ],units = NA n.fshk : [1 1 1 1 1 1 ],units = NA n.nshk : [1 1 1 1 1 1 ],units = NA var.fshk : [1 1 1 1 1 1 ],units = NA var.nshk : [1 1 1 1 1 1 ],units = NA catch.n : [1 1 1 1 1 1 ],units = NA stock.n : [10 61 1 1 1 1 ],units = NA harvest : [10 61 1 1 1 1 ],units = NA : index : [9 11 ] index.res : [9 11 ] index.hat : [9 11 ] index.var : [9 11 ] : Error in object@index[[i]]: attempt to select less than one element in integerOneIndex

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/flr/FLXSA/issues/6?email_source=notifications&email_token=AAM2VJ4IVXY2FKNGPFO3A7DRGCWNBA5CNFSM4IOCPJT2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEOAPNKY#issuecomment-595654315, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAM2VJ4VWBJMIQG7IYT2VL3RGCWNBANCNFSM4IOCPJTQ.

iagomosqueira commented 4 years ago

I still do not get any error when run with ple4 or your dataset when calling FLXSA

res <- FLXSA(xsa.stock,xsa.indices,control=xsa.control)
res <- FLXSA(xsa.stock,xsa.indices,control=xsa.control, TRUE)

If you do

.Call("runFLXSA", xsa.stock, xsa.indices, xsa.control, FALSE)
.Call("runFLXSA", xsa.stock, xsa.indices, xsa.control, TRUE)

then there is an error from summary(FLXSA) because the index FLQuants returned by .Call() does not have names, so summary fails. The names are assigned inside FLXSA() after .Call() returns the result.

But the error I get is

Error in object@index[[i]] :
  attempt to select less than one element in integerOneIndex

What happens if you try

res <- .Call("runFLXSA", xsa.stock, xsa.indices, xsa.control, TRUE)

Can you also try openning R from the command line with

R --vanilla

to see if any other package is interfering with [[ for FLlst?

iagomosqueira commented 4 years ago

Sorry, just saw your testing code, let me try with that

iagomosqueira commented 4 years ago

With FLCore, FLAssess and FLXSA from the FLR repository, I get

> test(ple4, ple4.indices[1:3], control, "FLXSA")
test(ple4, ple4.indices[1:3], control, ".Call")
 [1] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
> test(ple4, ple4.indices[1:3], control, ".Call")
 [1] TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE

I get the same result in Linux, R 2.6.3, and Windows 10, R 2.6.1.

iagomosqueira commented 4 years ago

Tested in RStudio Version 1.2.5033 in Linux with no problem.

colinpmillar commented 4 years ago

Yeah, me too. Tested in the latest rocker/tidyverse container with rstudio 1.2.5.033

docker run --rm -p 127.0.0.1:8787:8787 -e DISABLE_AUTH=true rocker/tidyverse

with no problems.

colinpmillar commented 4 years ago

So it seems to be windows only, and only when you run FLXSA in rstudio in windows and radian (https://github.com/randy3k/radian) in windows. I suspect that radian in linux will not have the issue, though I don't know.

This is well beyond my knowledge, so my only, pretty lame suggestion is to wrap the .Call("runFLXSA", ...) in a function

runFLXSA <- function(stock, indices, diag.flag, max.tries = 50) {
  res <- try(.Call("runFLXSA", stock, indices, diag.flag))
  i <- 0
  while (!is.FLXSA(res) && i < max.tries) {
    i <- i + 1
    res <- try(.Call("runFLXSA", stock, indices, diag.flag))
  }
  res
}

you could have a Sys.info()[["sysname"]] == "Windows" in there if you only want to do the try(...) and while(...) {...} when running on windows.

colinpmillar commented 4 years ago

tested and working in radian in linux container

docker run -it continuumio/miniconda3 /bin/bash
# https://docs.anaconda.com/anaconda/user-guide/tasks/using-r-language/
conda install r-base
pip install -U radian
radian

now in radian console:

install.packages("iterators", repos = "https://cloud.r-project.org/")
install.packages(c("MASS", "lattice", "Matrix"), repos = "https://cloud.r-project.org/")
install.packages(c("FLCore", "FLAssess", "FLXSA"), repos = "http://flr-project.org/R")
library(FLXSA)
data(ple4)
data(ple4.indices)
control <- FLXSA.control()
res <- .Call("runFLXSA", ple4, ple4.indices[1], control, TRUE)
iagomosqueira commented 4 years ago

OK, managed to replicate it, Windows 10, R 3.6.1 RStudio 1.2. And yes, the problem appears to be on the object being returned by .Call() being broken. How could this happen on just this setup I don't know.

The indices object is of length 3, but all 3 FLIndex elements are valid. Running with one of nthem each time still returns the same error.

I have tried @laurieKell suggestion to save or dump the object, but by the time it gets to fqs() it is already invalid and the Rdata cannot be loaded. Any ideas?

zenn1989 commented 4 years ago

Seems that @laurieKell is right about cpp code issue. Error happens only in Windows with Rstudio 1.2.xx (1.1.xx still working great). Bug happens when you call c++ : 

obj <- .Call("runFLXSA", stock, indices, control, diag.flag)

and the next call to obj will throw exception: 

negative length vectors are not allowed

any actions with obj (save or print) will finished with this error.

Look's like we need somebody who got c++ skills to review flxsa.cpp sources ...

iagomosqueira commented 3 years ago

Closing this as it seems to be limited to certain setups. There is little chance of anybody taking this on board. Effort is now placed in other stock assessment models, like FLa4a.