[Rd] possible bug in summaryRprof with memory.profiling=TRUE

Kasper Daniel Hansen kasperdanielhansen at gmail.com
Wed Oct 6 15:31:52 CEST 2010


I apologize for the somewhat brief bug report, but we are getting
close to the release of 2.12 and I thought it was better to send
something brief than to stay silent.

R version 2.12.0 beta (2010-10-06 r53205)

compiled with memory-profiling enabled.  I have very little experience
with memory profiling, but the following works
  Rprof(memory.profiling = TRUE)
  example(glm)
  Rprof(NULL)
  summaryRprof(memory = "stats")

Now, I am working on a personal function involving a number of nested functions,
  smoothP contains smoothChromosome contains smoothSample
These functions contains a number of print(gc()) calls that I have
left untouched (not sure if this is related).  My code also used
multicore, but currently it uses mclapply(..., mc.cores = 1) which (as
far as I can see) simply calls lapply.

When I run
  Rprof(memory.profiling = TRUE)
  BSdata <- smoothP(BSdata.tmp, ns = 70, h = 1000, mc.cores = 1)
  Rprof(NULL)
  summaryRprof(memory = "stats")

I get an error:
 Error in tapply(1L:961L, list(index = c("\"smoothP\":\"print\"",
"\"smoothP\":\"print\"",  :
  arguments must have same length

Enter a frame number, or 0 to exit

1: summaryRprof(memory = "stats")
2: Rprof_memory_summary(filename = con, chunksize = chunksize, aggregate = ind
3: by(memcounts, index, function(these) with(these, round(c(vsize.small = mean
4: by.data.frame(memcounts, index, function(these) with(these, round(c(vsize.s
5: eval(substitute(tapply(1:nd, IND, FUNx, simplify = simplify)), data)
6: eval(expr, envir, enclos)
7: tapply(1:961, list(index = c("\"smoothP\":\"print\"",
"\"smoothP\":\"print\

Selection: 2
Called from: eval(expr, envir, enclos)
Browse[1]> dim(memcounts)
[1] 961   4
Browse[1]> length(index)
[1] 5961
Browse[1]> head(index)
[1] "\"smoothP\":\"print\"" "\"smoothP\":\"print\""
"\"smoothP\":\"print\""
[4] "\"smoothP\":\"print\"" "\"smoothP\":\"print\""
"\"smoothP\":\"print\""
Browse[1]> tail(index)
[1] "\"smoothP\":\"print\"" "\"smoothP\":\"print\""
"\"smoothP\":\"print\""
[4] "\"smoothP\":\"print\"" "\"smoothP\":\"list\""
"\"smoothP\":\"list\""
Browse[1]>

The file Rprof.out has 5962 lines according to wc -l, and looks
normal.  The file is available at
  http://www.biostat.jhsph.edu/~khansen/Rprof.out
It parses correctly if I just do
  summaryRprof()
(without the memory = "stats" option).  If I use memory = "tseries" I
also get a row number mismatch:

Error in data.frame(..., check.names = FALSE) :
  arguments imply differing number of rows: 961, 5961

Enter a frame number, or 0 to exit

1: summaryRprof("http://www.biostat.jhsph.edu/~khansen/Rprof.out", memory = "t
2: Rprof_memory_summary(filename = con, chunksize = chunksize, label = index,
3: cbind(memcounts, labels)
4: cbind(deparse.level, ...)
5: data.frame(..., check.names = FALSE)

Selection:

I see this in R-devel (build today) as well.

Kasper



More information about the R-devel mailing list