MichaelChirico
commented
4 years ago Oups, tbl <- 1:4 and not 1:5, sorry
METADATA
- Comment author - Emmanuel CURIS
- Timestamp - 2019-05-28 08:56:26 UTC
Closed MichaelChirico closed 4 years ago
MichaelChirico
commented
4 years ago Oups, tbl <- 1:4 and not 1:5, sorry
MichaelChirico
commented
4 years ago tbl <- 1:4 names( tbl ) <- c( "OK", "Correct", "With é", "OK" ) pie( tbl )
works fine on my Debian Linux system.
-k
MichaelChirico
commented
4 years ago (In reply to Kurt Hornik from comment #2)
tbl <- 1:4
names( tbl ) <- c( "OK", "Correct", "With é", "OK" )
pie( tbl )
works fine on my Debian Linux system.
-kMakes things difficult to identify... I am ready to give any needed information on the system/libraries that could be needed to identify the problem.
For instance :
sessionInfo()R version 3.6.0 (2019-04-26) Platform: x86_64-pc-linux-gnu (64-bit) Running under: Mageia 6
Matrix products: default BLAS: /usr/local/lib64/R/lib/libRblas.so LAPACK: /usr/local/lib64/R/lib/libRlapack.so
locale:
[1] LC_CTYPE=fr_FR.UTF-8 LC_NUMERIC=C
[3] LC_TIME=fr_FR.UTF-8 LC_COLLATE=fr_FR.UTF-8
[5] LC_MONETARY=fr_FR.UTF-8 LC_MESSAGES=fr_FR.UTF-8
[7] LC_PAPER=fr_FR.UTF-8 LC_NAME=C
[9] LC_ADDRESS=C LC_TELEPHONE=C
[11] LC_MEASUREMENT=fr_FR.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_3.6.0
$ uname -a Linux [...) 4.18.20-server-1.mga6 #1 SMP Sun Nov 25 18:16:20 UTC 2018 x86_64 x86_64 x86_64 GNU/Linux
$ xdpyinfo name of display: :0.0 version number: 11.0 vendor string: Mageia vendor release number: 11905000 maximum request size: 16777212 bytes
MichaelChirico
commented
4 years ago In one line, this is
pie(c(OK = 1, Correct = 2, "With é" = 3, "OK, too" = 4))and that works nicely (and does nicely "plot" the 'é' as well), also for me (and most probably almost all R users ..)
sessionInfo()R version 3.6.0 (2019-04-26) Platform: x86_64-pc-linux-gnu (64-bit) Running under: Fedora 28 (Twenty Eight)
Matrix products: default BLAS: /...../R-3.6.0-inst/lib/libRblas.so LAPACK: /...../R-3.6.0-inst/lib/libRlapack.so
locale:
[1] LC_CTYPE=de_CH.UTF-8 LC_NUMERIC=C LC_TIME=en_US.UTF-8
[4] LC_COLLATE=de_CH.UTF-8 LC_MONETARY=en_US.UTF-8 LC_MESSAGES=de_CH.UTF-8
[7] LC_PAPER=de_CH.UTF-8 LC_NAME=C LC_ADDRESS=C
[10] LC_TELEPHONE=C LC_MEASUREMENT=de_CH.UTF-8 LC_IDENTIFICATION=C
attached base packages: [1] graphics grDevices datasets stats utils methods base
other attached packages: [1] fortunes_1.5-4 sfsmisc_1.1-4
loaded via a namespace (and not attached): [1] compiler_3.6.0 tools_3.6.0
capabilities() jpeg png tiff tcltk X11 aqua http/ftp
TRUE TRUE TRUE TRUE TRUE FALSE TRUE
sockets libxml fifo cledit iconv NLS profmem
TRUE TRUE TRUE FALSE TRUE TRUE FALSE
cairo ICU long.double libcurl
TRUE TRUE TRUE TRUE pie( c(OK = 1, Correct = 2, "With é" = 3, "OK, too" = 4))
.Device[1] "X11cairo"
--------
(In reply to Emmanuel CURIS from comment #0)
The bug is still present in R 3.6.0 (but this version is not proposed in the
version selector above...)But the version 3.6.x is proposed ... I had decided we would not want so extremely many versions...
MichaelChirico
commented
4 years ago As you are on a Linux, can you please run R as
R -d gdb --vanilla
and then press 'r' (= 'run') to start R, then at the prompt do the
pie(c(OK = 1, Correct = 2, "With �" = 3, "OK, too" = 4))
and when it "loops" you should be able to interrupt (only when inside debugger) and then type 'bt' to see the back trace etc
MichaelChirico
commented
4 years ago (In reply to Martin Maechler from comment #4 and #5) Sorry for missing the 3.6.x, I relied on the alphabetical order so didn't tried to go upward in the list, I missed the change from R 3.5.0 to 3.6.x without R :(
Results from R functions:
capabilities() jpeg png tiff tcltk X11 aqua
TRUE TRUE TRUE TRUE TRUE FALSE http/ftp sockets libxml fifo cledit iconv TRUE TRUE TRUE TRUE TRUE TRUE NLS profmem cairo ICU long.double libcurl TRUE FALSE TRUE TRUE TRUE TRUE
.Device[1] "null device"
pie(c(OK = 1, Correct = 2, "With é" = 3, "OK, too" = 4))^C
.Device[1] "X11"
May be the difference is between cairo and native X11 ?
Results from the call from gdb
pie(c(OK = 1, Correct = 2, "With é" = 3, "OK, too" = 4))^C Program received signal SIGINT, Interrupt. 0x0000000000511077 in CONS_NR (car=0x842ef8, cdr=cdr@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x841010) at memory.c:2411 2411 QUICK_GET_FREE_NODE(s);
(gdb) bt
at memory.c:2411rho=rho@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x32c27b8, useCache=useCache@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=TRUE) at eval.c:6837at eval.c:620newrho=newrho@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x32c27b8, sysparent=<optimized out>,
rho=rho@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x873680, arglist=arglist@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x32c2828,
op=op@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x179d1b8) at eval.c:1780op=op@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x179d1b8, arglist=<optimized out>, rho=rho@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x873680,
suppliedvars=<optimized out>) at eval.c:1706at envir.c:3618family=<optimized out>) at devX11.c:1778ascent=0x7fffffff9810, descent=0x7fffffff9818, width=0x7fffffff9820,
dd=<optimized out>) at devX11.c:1836str=str@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x14aa4d8 "With �", enc=CE_UTF8, xc=xc@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=1,
yc=<optimized out>, yc@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=nan(0x0000007a2), rot=rot@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0,
gc=gc@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x7fffffff98a0, dd=dd@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x288b190) at engine.c:1804y=250.30229443143043, coords=coords@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=13,
str=str@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x14aa4d8 "With �", enc=CE_UTF8, xc=xc@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=1,
yc=yc@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=nan(0x0000007a2), rot=rot@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0, dd=dd@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x288b190)
at graphics.c:3023op=op@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x85e308, args=args@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x2973ff8, env=env@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x2974a78)
at dotcode.c:548args=0x2973ff8, env=0x2974a78) at dotcode.c:1318rho=rho@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x2974a78, useCache=useCache@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=TRUE) at eval.c:6722at eval.c:620newrho=newrho@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x2974a78, sysparent=<optimized out>,
rho=rho@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x2971268, arglist=arglist@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x2971578,
op=op@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x293e898) at eval.c:1780arglist=<optimized out>, rho=0x2971268, suppliedvars=<optimized out>)
at eval.c:1706op=op@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x293e898, args=<optimized out>, rho=rho@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x2971268,
newvars=newvars@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x29710a8) at objects.c:118sxp=0x293e898, dotClass=0x841010, cptr=cptr@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x7fffffffbbb0,
method=method@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x12c5ec0, generic=generic@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x9ad518 "text",
rho=0x2971268, callrho=0x27ee040, defrho=0x13e4138) at objects.c:436obj=obj@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x296b9d8, call=call@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x11ac228,
args=args@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x841010, rho=rho@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x2971268,
callrho=callrho@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x27ee040, defrho=0x13e4138, ans=0x7fffffffad18)
at objects.c:486args=<optimized out>, env=0x2971268) at objects.c:565rho=rho@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x2971268, useCache=useCache@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=TRUE) at eval.c:6785at eval.c:620newrho=newrho@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x2971268, sysparent=<optimized out>,
rho=rho@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x27ee040, arglist=arglist@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x2971578,
op=op@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x11abe00) at eval.c:1780op=op@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x11abe00, arglist=<optimized out>, rho=rho@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x27ee040,
suppliedvars=<optimized out>) at eval.c:1706rho=rho@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x27ee040, useCache=useCache@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=TRUE) at eval.c:6733at eval.c:620newrho=newrho@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x27ee040, sysparent=<optimized out>,
rho=rho@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x873680, arglist=arglist@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x27ee350,
op=op@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x2706898) at eval.c:1780op=op@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x2706898, arglist=<optimized out>, rho=rho@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x873680,
suppliedvars=<optimized out>) at eval.c:1706at eval.c:743savestack=<optimized out>, browselevel=0, state=0x7fffffffcc70)
at main.c:260browselevel=0) at main.c:310at Rmain.c:29(gdb) cont Continuing. ^C Program received signal SIGINT, Interrupt. R_Newhashpjw (s=s@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x7ffff24989b9 ".X11env") at envir.c:222 222 for (p = (char ) s; p; p++) { & (gdb) cont Continuing. ^C Program received signal SIGINT, Interrupt. 0x00000000004cb4f6 in bcEval (body=body@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x179d148, rho=rho@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=0x2b2dfd8, useCache=useCache@<::CENSORED -- SEE ORIGINAL ON BUGZILLA::>=TRUE) at eval.c:6517 6517 OP(GETVAR, 1): DO_GETVAR(FALSE, FALSE);
MichaelChirico
commented
4 years ago Indeed, your .Device |--> "X11" has been the clue.
Now I can replicate:
X11(type = "Xlib") .Device # now shows "X11" pie(c(OK = 1, Correct = 2, "With é" = 3, "OK, too" = 4)) # --> hangs !!
-- and your backtrace is revealing too: I think it "shows" that the font metrics somehow can't be found for the non-ASCII character there ...
MichaelChirico
commented
4 years ago (In reply to Martin Maechler from comment #7)
Indeed, your .Device |--> "X11" has been the clue.
Now I *can* replicate:
X11(type = "Xlib")
.Device # now shows "X11"
pie(c(OK = 1, Correct = 2, "With é" = 3, "OK, too" = 4)) # --> hangs !!
-- and your backtrace is revealing too: I think it "shows" that the font
metrics somehow can't be found for the non-ASCII character there ...Great (in some sens at least). What is quite strange is that I never experienced such a thing with other graphical functions in R, despite a large use of UTF-8 characters in my scripts. But I didn't managed to understand what pie() does differently than the others...
MichaelChirico
commented
4 years ago (In reply to Emmanuel CURIS from comment #8)
What is quite strange is that I never experienced such a thing with other
graphical functions in R, despite a large use of UTF-8 characters in my
scripts. But I didn't managed to understand what pie() does differently
than the others...Well, it just calls text().
I also get the infinite loop
X11(type = "Xlib") plot(1,1) ; text(1,1, "with ä")
Interestingly, it seems to only happen when calling text() directly, but not e.g. here :
plot(1,1, xlab = "with é", main = "Voilà")
and here
mtext(side=3, "¡ Voilà, Martin Mächler in Zürich !")
but indeed "hangs" with text()
So the font problem only happens when going via text() but not when going via title() {which is used for the above 'xlab = , main = ') or via mtext().
Indeed, these R functions each call into different C code,
C_text(), vs C_title() vs C_mtext()
and it seems C_text() does something not as well as the other two.
All three function's source code is in
Maybe Paul can have a look?
MichaelChirico
commented
4 years ago (In reply to Martin Maechler from comment #9)
(In reply to Emmanuel CURIS from comment #8)
> What is quite strange is that I never experienced such a thing with other
> graphical functions in R, despite a large use of UTF-8 characters in my
> scripts. But I didn't managed to understand what pie() does differently
> than the others...
Well, it just calls text().
I also get the infinite loop
X11(type = "Xlib")
plot(1,1) ; text(1,1, "with ä")
Indeed, these R functions each call into different C code,
C_text(), vs C_title() vs C_mtext()
and it seems C_text() does something not as well as the other two.
All three function's source code is in <Rsrc>/src/library/graphics/src/plot.c
Maybe Paul can have a look?It seems to have something to do with adjustment values, and the default:
plot( 1, 1 ) text(1,1, "with ä", adj = c( 0, 0 ) ) # Does not hang text(1,1, "with ä", adj = c( 0, 1 ) ) # Does not hang text(1,1, "with ä", adj = c( 1, 0 ) ) # Does not hang text(1,1, "with ä", adj = c( 1, 1 ) ) # Does not hang text(1,1, "with ä", adj = c( 0.5, 0.5 ) ) # Does not hang
text(1,1, "with ä", adj = NULL ) # DOES HANG
MichaelChirico
commented
4 years ago The special code for handling the NULL adjustment in plot.c seems to be
PROTECT(adj = CAR(args));
if (isNull(adj) || (isNumeric(adj) && length(adj) == 0)) {
adjx = gpptr(dd)->adj;
adjy = NA_REAL;
}and the NA_REAL may be the cause since
plot( 1, 1 )
text( 1, 1, "é", adj = c( 1, NA ) )does also hang, but
text( 1, 1, "é", adj = c( NA, 1 ) )
does...
MichaelChirico
commented
4 years ago (does not, for the second, sorry)
MichaelChirico
commented
4 years ago Thanks for the report, fixed in R-devel 77638.
Hi, I've just noticed that calling pie() on a table whose name contains UTF-8 characters seems to make pie() enter in a infinite loop.
Just try the following code:
tbl <- c( 1, 2, 3, 4 5) names( tbl ) <- c( "OK", "Correct", "With é", "OK" ) pie( tbl )
The display stops when trying to display the third label. It works perfectly when removing the é, and there is no problem when using barplot()
The bug is still present in R 3.6.0 (but this version is not proposed in the version selector above...)
Tried on linux/Mageia, with a fresh installed R 3.6.0 and no other package loaded...
METADATA