Hi,
I'm running into a case where byte-compilation changes
the semantic of a function. This is with R 3.4.0:
foo <- function(x) { TRUE && x }
foo(c(a=FALSE))
# [1] FALSE # OK
foo(c(a=TRUE))
# [1] TRUE # OK
foo(c(a=FALSE))
# a # ????
# FALSE
The 3rd call returned a result that it different from the 1st
call!
After scratching my head for a while, I found out that there
is a lot going on:
1) When calling foo the first 2 times, the function is not
byte-compiled so is behaving as expected.
2) However after the 2nd call, the function gets automatically
compiled. This seems to be a new feature in R 3.4.0 but
you'll have to grep the NEWS file to find out (search for
JIT). The man page for cmpfun/enableJIT nicely explains
the JIT levels but I couldn't find anywhere that the level
is set to 3 by default. I don't even know how one is
supposed to get the current level. Calling enableJIT() to
change the level returns a value, and this value might
look like the previous level, but this is undocumented and
it returns 0 instead of 3 the 1st time I call it anyway.
3) There is some bug in the byte-compiler that seems to make it
treat `&&` like if it were `&` when compiling foo.
Cheers,
H.
> sessionInfo()
R version 3.4.0 (2017-04-21)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: Ubuntu 16.04.2 LTS
Matrix products: default
BLAS: /home/biocbuild/bbs-3.5-bioc/R/lib/libRblas.so
LAPACK: /home/biocbuild/bbs-3.5-bioc/R/lib/libRlapack.so
locale:
[1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C
[3] LC_TIME=en_US.UTF-8 LC_COLLATE=en_US.UTF-8
[5] LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8
[7] LC_PAPER=en_US.UTF-8 LC_NAME=C
[9] LC_ADDRESS=C LC_TELEPHONE=C
[11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C
attached base packages:
[1] stats graphics grDevices utils datasets methods base
loaded via a namespace (and not attached):
[1] compiler_3.4.0
--
Herv? Pag?s
Program in Computational Biology
Division of Public Health Sciences
Fred Hutchinson Cancer Research Center
1100 Fairview Ave. N, M1-B514
P.O. Box 19024
Seattle, WA 98109-1024
E-mail: hpages at fredhutch.org
Phone: (206) 667-5791
Fax: (206) 667-1319
On 26/04/2017 4:36 PM, Herv? Pag?s wrote:> Hi, > > I'm running into a case where byte-compilation changes > the semantic of a function. This is with R 3.4.0: > > foo <- function(x) { TRUE && x } > > foo(c(a=FALSE)) > # [1] FALSE # OK > > foo(c(a=TRUE)) > # [1] TRUE # OK > > foo(c(a=FALSE)) > # a # ???? > # FALSE > > The 3rd call returned a result that it different from the 1st > call!That does look like a bug. Please file a bug report.> > After scratching my head for a while, I found out that there > is a lot going on: > > 1) When calling foo the first 2 times, the function is not > byte-compiled so is behaving as expected. > > 2) However after the 2nd call, the function gets automatically > compiled. This seems to be a new feature in R 3.4.0 but > you'll have to grep the NEWS file to find out (search for > JIT).The NEWS file is where we usually report news, so this isn't really that much of a surprise. If you don't want to read it all at once, you can see it as it changes on the RSS feed at http://developer.r-project.org/RSSfeeds.html The man page for cmpfun/enableJIT nicely explains> the JIT levels but I couldn't find anywhere that the level > is set to 3 by default. I don't even know how one is > supposed to get the current level. Calling enableJIT() to > change the level returns a value, and this value might > look like the previous level, but this is undocumented and > it returns 0 instead of 3 the 1st time I call it anyway. > > 3) There is some bug in the byte-compiler that seems to make it > treat `&&` like if it were `&` when compiling foo.It's a bug, but that doesn't describe it completely. TRUE & c(a=TRUE, b=FALSE) doesn't give the same thing as foo(c(a=TRUE, b=FALSE)). Duncan Murdoch> > Cheers, > H. > > > sessionInfo() > R version 3.4.0 (2017-04-21) > Platform: x86_64-pc-linux-gnu (64-bit) > Running under: Ubuntu 16.04.2 LTS > > Matrix products: default > BLAS: /home/biocbuild/bbs-3.5-bioc/R/lib/libRblas.so > LAPACK: /home/biocbuild/bbs-3.5-bioc/R/lib/libRlapack.so > > locale: > [1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C > [3] LC_TIME=en_US.UTF-8 LC_COLLATE=en_US.UTF-8 > [5] LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8 > [7] LC_PAPER=en_US.UTF-8 LC_NAME=C > [9] LC_ADDRESS=C LC_TELEPHONE=C > [11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C > > attached base packages: > [1] stats graphics grDevices utils datasets methods base > > loaded via a namespace (and not attached): > [1] compiler_3.4.0 >
On 04/26/2017 01:51 PM, Duncan Murdoch wrote:> On 26/04/2017 4:36 PM, Herv? Pag?s wrote: >> Hi, >> >> I'm running into a case where byte-compilation changes >> the semantic of a function. This is with R 3.4.0: >> >> foo <- function(x) { TRUE && x } >> >> foo(c(a=FALSE)) >> # [1] FALSE # OK >> >> foo(c(a=TRUE)) >> # [1] TRUE # OK >> >> foo(c(a=FALSE)) >> # a # ???? >> # FALSE >> >> The 3rd call returned a result that it different from the 1st >> call! > > That does look like a bug. Please file a bug report. > >> >> After scratching my head for a while, I found out that there >> is a lot going on: >> >> 1) When calling foo the first 2 times, the function is not >> byte-compiled so is behaving as expected. >> >> 2) However after the 2nd call, the function gets automatically >> compiled. This seems to be a new feature in R 3.4.0 but >> you'll have to grep the NEWS file to find out (search for >> JIT). > > The NEWS file is where we usually report news, so this isn't really that > much of a surprise. If you don't want to read it all at once, you can > see it as it changes on the RSS feed at > > https://urldefense.proofpoint.com/v2/url?u=http-3A__developer.r-2Dproject.org_RSSfeeds.html&d=DwIGaQ&c=eRAMFD45gAfqt84VtBcfhQ&r=BK7q3XeAvimeWdGbWY_wJYbW0WYiZvSXAJJKaaPhzWA&m=2Qn1jcR4dORPq3qVhYNnz_O64UneKth6ebbeC4CEtRI&s=O1JVjaNkaGIrMcR0qo58R2OrzIk0nOMoL89tskLXeK0&e>The surprise here is that this feature is not documented in TFM. If you're trying to understand a feature and you don't know it's a new feature, NEWS is not necessarily the 1st place you go. NEWS is not TFM. 'svn log' in not TFM. H.> > The man page for cmpfun/enableJIT nicely explains >> the JIT levels but I couldn't find anywhere that the level >> is set to 3 by default. I don't even know how one is >> supposed to get the current level. Calling enableJIT() to >> change the level returns a value, and this value might >> look like the previous level, but this is undocumented and >> it returns 0 instead of 3 the 1st time I call it anyway. >> >> 3) There is some bug in the byte-compiler that seems to make it >> treat `&&` like if it were `&` when compiling foo. > > It's a bug, but that doesn't describe it completely. TRUE & c(a=TRUE, > b=FALSE) doesn't give the same thing as foo(c(a=TRUE, b=FALSE)). > > Duncan Murdoch > >> >> Cheers, >> H. >> >> > sessionInfo() >> R version 3.4.0 (2017-04-21) >> Platform: x86_64-pc-linux-gnu (64-bit) >> Running under: Ubuntu 16.04.2 LTS >> >> Matrix products: default >> BLAS: /home/biocbuild/bbs-3.5-bioc/R/lib/libRblas.so >> LAPACK: /home/biocbuild/bbs-3.5-bioc/R/lib/libRlapack.so >> >> locale: >> [1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C >> [3] LC_TIME=en_US.UTF-8 LC_COLLATE=en_US.UTF-8 >> [5] LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8 >> [7] LC_PAPER=en_US.UTF-8 LC_NAME=C >> [9] LC_ADDRESS=C LC_TELEPHONE=C >> [11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C >> >> attached base packages: >> [1] stats graphics grDevices utils datasets methods base >> >> loaded via a namespace (and not attached): >> [1] compiler_3.4.0 >> > > ______________________________________________ > R-devel at r-project.org mailing list > https://urldefense.proofpoint.com/v2/url?u=https-3A__stat.ethz.ch_mailman_listinfo_r-2Ddevel&d=DwIGaQ&c=eRAMFD45gAfqt84VtBcfhQ&r=BK7q3XeAvimeWdGbWY_wJYbW0WYiZvSXAJJKaaPhzWA&m=2Qn1jcR4dORPq3qVhYNnz_O64UneKth6ebbeC4CEtRI&s=pXXxO1dFT00kYhhUuO35VRCEYFiw6n2Sz4nSMb6OdxQ&e>-- Herv? Pag?s Program in Computational Biology Division of Public Health Sciences Fred Hutchinson Cancer Research Center 1100 Fairview Ave. N, M1-B514 P.O. Box 19024 Seattle, WA 98109-1024 E-mail: hpages at fredhutch.org Phone: (206) 667-5791 Fax: (206) 667-1319
Thanks for the report. Fixed in R_devel with r72631; I'll port to R-patched later today or tomorrow. The default setting and how to get the current settings are documented in the Details section of the enableJIT help page at the end of the paragraph for enableJIT. Best, luke On Wed, 26 Apr 2017, Herv? Pag?s wrote:> Hi, > > I'm running into a case where byte-compilation changes > the semantic of a function. This is with R 3.4.0: > > foo <- function(x) { TRUE && x } > > foo(c(a=FALSE)) > # [1] FALSE # OK > > foo(c(a=TRUE)) > # [1] TRUE # OK > > foo(c(a=FALSE)) > # a # ???? > # FALSE > > The 3rd call returned a result that it different from the 1st > call! > > After scratching my head for a while, I found out that there > is a lot going on: > > 1) When calling foo the first 2 times, the function is not > byte-compiled so is behaving as expected. > > 2) However after the 2nd call, the function gets automatically > compiled. This seems to be a new feature in R 3.4.0 but > you'll have to grep the NEWS file to find out (search for > JIT). The man page for cmpfun/enableJIT nicely explains > the JIT levels but I couldn't find anywhere that the level > is set to 3 by default. I don't even know how one is > supposed to get the current level. Calling enableJIT() to > change the level returns a value, and this value might > look like the previous level, but this is undocumented and > it returns 0 instead of 3 the 1st time I call it anyway. > > 3) There is some bug in the byte-compiler that seems to make it > treat `&&` like if it were `&` when compiling foo. > > Cheers, > H. > > > sessionInfo() > R version 3.4.0 (2017-04-21) > Platform: x86_64-pc-linux-gnu (64-bit) > Running under: Ubuntu 16.04.2 LTS > > Matrix products: default > BLAS: /home/biocbuild/bbs-3.5-bioc/R/lib/libRblas.so > LAPACK: /home/biocbuild/bbs-3.5-bioc/R/lib/libRlapack.so > > locale: > [1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C > [3] LC_TIME=en_US.UTF-8 LC_COLLATE=en_US.UTF-8 > [5] LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8 > [7] LC_PAPER=en_US.UTF-8 LC_NAME=C > [9] LC_ADDRESS=C LC_TELEPHONE=C > [11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C > > attached base packages: > [1] stats graphics grDevices utils datasets methods base > > loaded via a namespace (and not attached): > [1] compiler_3.4.0 > >-- Luke Tierney Ralph E. Wareham Professor of Mathematical Sciences University of Iowa Phone: 319-335-3386 Department of Statistics and Fax: 319-335-3017 Actuarial Science 241 Schaeffer Hall email: luke-tierney at uiowa.edu Iowa City, IA 52242 WWW: http://www.stat.uiowa.edu
On 04/26/2017 04:46 PM, luke-tierney at uiowa.edu wrote:> Thanks for the report. Fixed in R_devel with r72631; I'll port to > R-patched later today or tomorrow.Thanks!> > The default setting and how to get the current settings are documented > in the Details section of the enableJIT help page at the end of the > paragraph for enableJIT.Indeed. Sorry for missing that (I was playing with several versions of R and probably looked at the enableJIT man page in the wrong R). H.> > Best, > > luke > > On Wed, 26 Apr 2017, Herv? Pag?s wrote: > >> Hi, >> >> I'm running into a case where byte-compilation changes >> the semantic of a function. This is with R 3.4.0: >> >> foo <- function(x) { TRUE && x } >> >> foo(c(a=FALSE)) >> # [1] FALSE # OK >> >> foo(c(a=TRUE)) >> # [1] TRUE # OK >> >> foo(c(a=FALSE)) >> # a # ???? >> # FALSE >> >> The 3rd call returned a result that it different from the 1st >> call! >> >> After scratching my head for a while, I found out that there >> is a lot going on: >> >> 1) When calling foo the first 2 times, the function is not >> byte-compiled so is behaving as expected. >> >> 2) However after the 2nd call, the function gets automatically >> compiled. This seems to be a new feature in R 3.4.0 but >> you'll have to grep the NEWS file to find out (search for >> JIT). The man page for cmpfun/enableJIT nicely explains >> the JIT levels but I couldn't find anywhere that the level >> is set to 3 by default. I don't even know how one is >> supposed to get the current level. Calling enableJIT() to >> change the level returns a value, and this value might >> look like the previous level, but this is undocumented and >> it returns 0 instead of 3 the 1st time I call it anyway. >> >> 3) There is some bug in the byte-compiler that seems to make it >> treat `&&` like if it were `&` when compiling foo. >> >> Cheers, >> H. >> >> > sessionInfo() >> R version 3.4.0 (2017-04-21) >> Platform: x86_64-pc-linux-gnu (64-bit) >> Running under: Ubuntu 16.04.2 LTS >> >> Matrix products: default >> BLAS: /home/biocbuild/bbs-3.5-bioc/R/lib/libRblas.so >> LAPACK: /home/biocbuild/bbs-3.5-bioc/R/lib/libRlapack.so >> >> locale: >> [1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C >> [3] LC_TIME=en_US.UTF-8 LC_COLLATE=en_US.UTF-8 >> [5] LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8 >> [7] LC_PAPER=en_US.UTF-8 LC_NAME=C >> [9] LC_ADDRESS=C LC_TELEPHONE=C >> [11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C >> >> attached base packages: >> [1] stats graphics grDevices utils datasets methods base >> >> loaded via a namespace (and not attached): >> [1] compiler_3.4.0 >> >> >-- Herv? Pag?s Program in Computational Biology Division of Public Health Sciences Fred Hutchinson Cancer Research Center 1100 Fairview Ave. N, M1-B514 P.O. Box 19024 Seattle, WA 98109-1024 E-mail: hpages at fredhutch.org Phone: (206) 667-5791 Fax: (206) 667-1319