Re: [Rd] calling R API functions after engine shutdown

[Lukas Stadler]

If this is not considered to be safe, then I guess my message can be seen as a 
heads-up:
If R_ReleaseObject ever does something that is not safe after the R engine has 
shut down, you will start to see crashes from packages like “minqa”.
In FastR, we ended up adding a global flag that disables parts of the native 
API because of this.

- Lukas

> On 21 Sep 2017, at 17:53, Tomas Kalibera  wrote:
> 
> 
> Calling R_ReleaseObject in a C++ destructor is not reliable - it can be 
> bypassed by a non-local return, such as an error. Generally in R one cannot 
> use C++ destructors reliably for anything that the R runtime wouldn't do on 
> its own in case of a non-local return.
> 
> A destructor that calls just UNPROTECT, in a way that balances out the 
> protection stack (e.g. Rcpp Shield), is safe because R runtime balances the 
> protection stack on non-local return. Destructors used in code that will 
> never call into any R API (such as in a third party library) are safe, 
> because the R runtime could not do non-local return. All other destructors 
> are a problem.
> 
> UNPROTECT will work even during R shutdown.
> 
> In some cases cleanup code that would be put in C++ destructors, if they were 
> safe with R, can instead be put into R finalizers.
> 
> Tomas
> 
> 
> 
> On 09/21/2017 04:41 PM, Lukas Stadler wrote:
>> Hi!
>> 
>> We’ve recently come across an example where a package (minqa) creates an 
>> Rcpp Function object in a static variable.
>> This causes R_ReleaseObject to be called by the destructor at a very late 
>> point in time - as part of the system exit function:
>> 
>> static Function cf("c");
>> 
>> I’m wondering if that is considered to be “safe”?
>> Is the R engine supposed to stay in a state where calls to API functions are 
>> valid, even after it has shut down?
>> It probably only ever happens with the ReleaseObject function, but even that 
>> could cause problems, e.g., with more elaborate refcounting schemes.
>> 
>> - Lukas
>> __
>> R-devel@r-project.org mailing list
>> https://urldefense.proofpoint.com/v2/url?u=https-3A__stat.ethz.ch_mailman_listinfo_r-2Ddevel&d=DwIDaQ&c=RoP1YumCXCgaWHvlZYR8PQcxBKCX5YTpkKY057SbK10&r=6DzVcyWX6yyD2r4eAolUQDWVUN_xc9UW6UdTto2_bao&m=YltTQ4tUmVFt23AnHzDGTPVCXmlUiOq1ayKJzhavePU&s=qibOCsqERij4Ymssp_hIfB78JDOA8v0jSYJ7ZTnCNO4&e=
>>  
> 
> 

__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel

Re: [Rd] Revert to R 3.2.x code of logicalSubscript in subscript.c?

[Radford Neal]

Suharto,

If you're interested in performance with subscripting, you might want
to look at pqR (pqR-project.org).  It has some substantial performance
improvements for subscripting over R Core versions.  This is
especially true for the current development version of pqR (probably
leading to a new release in about a month).

You can look at a somewhat-stable snapshot of recent pqR development at

  https://github.com/radfordneal/pqR/tree/05e32fa6

In particular, src/main/subscript.c might be of interest.

Note that you should read mods-dir/README if you want to build this,
and in particular, you need to run create-configure in the top-level
source directory first.

I modified your tests a bit, including producing versions using both
vectors of length 1e8 like you did (which will not fit in cache) and
vectors of length 1e5 (which will fit in at least the L3 cache).  I
ran tests on an Intel Skylake processor (E3-1270v5 @ 3.6GHz), using
gcc 7.2 with -O3 -march=native -mtune=native.

I got the following results with R-3.4.2 (with R_ENABLE_JIT=0, which
is slightly faster than using the JIT compiler):

R-3.4.2, LARGE VECTORS: 

  > N <- 1e8; R <- 5
  > #N <- 1e5; R <- 1000
  > 
  > x <- numeric(N)
  > i <- rep(FALSE, length(x))# no reycling
  > system.time(for (r in 1:R) a <- x[i])
 user  system elapsed 
0.296   0.000   0.297 
  > i <- FALSE# recycling
  > system.time(for (r in 1:R) a <- x[i])
 user  system elapsed 
0.416   0.000   0.418 
  > 
  > x <- numeric(N)
  > i <- rep(TRUE, length(x))# no reycling
  > system.time(for (r in 1:R) a <- x[i])
 user  system elapsed 
1.416   0.352   1.773 
  > i <- TRUE# recycling
  > system.time(for (r in 1:R) a <- x[i])
 user  system elapsed 
1.348   0.264   1.613 
  > 
  > x <- numeric(N)
  > system.time(for (r in 1:R) a <- x[-1])
 user  system elapsed 
1.516   0.376   1.895 
  > system.time(for (r in 1:R) a <- x[2:length(x)])
 user  system elapsed 
1.516   0.308   1.827 
  > 
  > v <- 2:length(x)
  > system.time(for (r in 1:R) a <- x[v])
 user  system elapsed 
1.416   0.268   1.689 

R-3.4.2, SMALL VECTORS: 

  > #N <- 1e8; R <- 5
  > N <- 1e5; R <- 1000
  > 
  > x <- numeric(N)
  > i <- rep(FALSE, length(x))# no reycling
  > system.time(for (r in 1:R) a <- x[i])
 user  system elapsed 
0.088   0.000   0.089 
  > i <- FALSE# recycling
  > system.time(for (r in 1:R) a <- x[i])
 user  system elapsed 
0.084   0.000   0.084 
  > 
  > x <- numeric(N)
  > i <- rep(TRUE, length(x))# no reycling
  > system.time(for (r in 1:R) a <- x[i])
 user  system elapsed 
0.492   0.020   0.515 
  > i <- TRUE# recycling
  > system.time(for (r in 1:R) a <- x[i])
 user  system elapsed 
0.408   0.008   0.420 
  > 
  > x <- numeric(N)
  > system.time(for (r in 1:R) a <- x[-1])
 user  system elapsed 
0.508   0.004   0.516 
  > system.time(for (r in 1:R) a <- x[2:length(x)])
 user  system elapsed 
0.464   0.008   0.473 
  > 
  > v <- 2:length(x)
  > system.time(for (r in 1:R) a <- x[v])
 user  system elapsed 
0.428   0.000   0.428 

Here are the results with the development version of pqR (uncompressed
pointers, no byte compilation):

pqR (devel), LARGE VECTORS:

  > N <- 1e8; R <- 5
  > #N <- 1e5; R <- 1000
  > 
  > x <- numeric(N)
  > i <- rep(FALSE, length(x))# no reycling
  > system.time(for (r in 1:R) a <- x[i])
 user  system elapsed 
0.192   0.000   0.193 
  > i <- FALSE# recycling
  > system.time(for (r in 1:R) a <- x[i])
 user  system elapsed 
0.436   0.000   0.434 
  > 
  > x <- numeric(N)
  > i <- rep(TRUE, length(x))# no reycling
  > system.time(for (r in 1:R) a <- x[i])
 user  system elapsed 
0.768   0.216   0.988 
  > i <- TRUE# recycling
  > system.time(for (r in 1:R) a <- x[i])
 user  system elapsed 
0.832   0.272   1.105 
  > 
  > x <- numeric(N)
  > system.time(for (r in 1:R) a <- x[-1])
 user  system elapsed 
0.280   0.156   0.435 
  > system.time(for (r in 1:R) a <- x[2:length(x)])
 user  system elapsed 
0.252   0.184   0.436 
  > 
  > v <- 2:length(x)
  > system.time(for (r in 1:R) a <- x[v])
 user  system elapsed 
0.828   0.168   0.998 

pqR (devel), SMALL VECTORS:

  > #N <- 1e8; R <- 5
  > N <- 1e5; R <- 1000
  > 
  > x <- numeric(N)
  > i <- rep(FALSE, length(x))# no reycling
  > system.time(for (r in 1:R) a <- x[i])
 user  system elapsed 
0.040   0.000   0.038 
  > i <- FALSE# recycling
  > system.time(for (r in 1:R) a <- x[i])
 user  system elapsed 
0.084   0.000   0.087 
  > 
  > x <- numeric(N)
  > i <- rep(TRUE, length(x))# no reycling
  > system.time(for (r in 1:R) a <- x[i])
 user  system elapsed 
0.156   0.036   0.192 
  > i <- TRUE# recycling
  > system.time(for (r in 1:R) a <- x[i])
 user  system elapsed 
0.184   0.012   0.195 
  > 
  > x <- numeric(N)
  > system.time(for (r in 1:R) a <- x[-1])
 user  system elapsed 
0.060   0.012   0.075 
  > system.time(for (r in