gdbsource error

[jlaake]

There is a closed post on this subject and while I have tried various suggestions, none have worked for me on windows. Hans Skaug is here in San Diego teaching a TMB class so I decided to explore this further and came up with the following solution.

For Windows, gdbsource calls

if (.Platform$OS.type == "windows") {
    return(.gdbsource.win(file, interactive))

In that function it calls system with cmd <- paste("gdb Rterm -x", gdbscript)

This then uses gdb.exe in Rtools/bin (first in path) which contains a gdb.exe and a gdb64.exe. Presumably the first is for R32 and the second for R64. I use R64 and this appears to be the source of the problem. I removed gdb.exe and renamed gdb64.exe to gdb.exe. That is all it took for it to work. I was unable to get the 32bit version to work however but no loss for me.

However, I also found that gdbsource did not work if you had used precompile(). Not sure why but you can always remove with clean=TRUE while debugging code.

A suggestion would be to include a switch in .gdbsource.win to use gdb.exe if run in R32 and gdb64.exe for R64 and then the user would not need to make any changes.

[skaug]

Let me take the good news first: gdbsource now seems to work on Windows (R 3.5.0 and Rtools 3.5.0.1). Even better is that flag="-O0 -g" is working, which gives much better debugging than "-O1 -g".

For installation under Windows I followed:

https://github.com/kaskr/adcomp/wiki/Windows-installation

On the negative side I find that interactive=FALSE, as in:

gdbsource("file.R",interactive=FALSE)

is not working well (but it has worked for me in previous versions of R/Rtools) For instance, when the R script contains a syntax error I get the confusing message

Error in line:length(x) : argument of length 0

which also occurs if gdb itself is not working properly

For windows using facing problems, I recommend to try gdbsource both with "interactive=TRUE" and "interactive=FALSE"

---

Details about version:

R.version
platform x86_64-w64-mingw32
arch x86_64
os mingw32
system x86_64, mingw32
status
major 3
minor 5.0

Rtools version 3.5.0.1

[kaskr]

@skaug I seem to recall that -O0 -g only works for toy models such as linreg.cpp whereas something more complex like sam.cpp fails. Has that changed with R 3.5.0 and Rtools 3.5.0.1 ?

[skaug]

Thanks. I had only tried on a toy model. sam.cpp still does not work. In fact, it will not even compile with -O0 -g.