[Bioc-devel] unexpected build and check error

[Martin Morgan]

On 04/04/2014 10:36 AM, Luo Weijun wrote:
> Dan,
> Thanks for the followup. That’s wired. I use R 3.0.0 here. It is different from R-3.1.0, but I don’t think it would make such a difference in generating the error below:
> Error in bods[idx, "id.type"] : subscript out of bounds
>
> I run the exact same build command on my local machine (as below), I got the same error indeed:
> R CMD build --keep-empty-dirs --no-resave-data gage
>
> But if I install gage package first, then then create vignettes manually, everything is fine.
> sudo R CMD INSTALL gage
> cd gage/vignettes/
> R CMD Sweave gage.Rnw
> R CMD texi2dvi -p gage.tex
>
> I am not sure, why there is such a difference. Do you have any idea on this?

I can reproduce this providing the path to the gage source tree

   tools::buildVignettes(dir="path/to/gage", quiet=FALSE)

you'll see that this builds the dataPrep vignette first, and then the gage 
vignette. It seems like dataPrep loads data bods from pathview into the global 
environment, and when the gage vignette is being processed it attempts to load 
an identically named data set into the (same) global environment.

The (partial) output of buildVignettes is in gage/vignettes/gage.tex; editing 
gage.Rnw to include a print(ls()) early in the vignette confirms that an object 
'bods' already exists.

It's surprising to me that R re-uses the same environment to build each 
vignette, and I'll take this up on the R-devel mailing list.

The data() function can arrange to load data into an arbitrary environment; for 
a function it seems better not to write to the global environment, and something 
like

   env = new.env(parent=emptyenv())
   obj = data("bods",envir=env)
   bods = env[[obj]]

is better.

Hopefully this helps to address your problem.

Martin

>
> I suspected that bods data file (bods.rda in the data directory) was not updated or have some problem somehow. Below is the problematic code chunk in the go.gset.R file:
>
> …
>        idx=which(bods[,"species"] == species)
>        if(length(idx)!=1) stop("bad species value")
>        pkg.name = bods[idx, "package"]
>        id.type=toupper(bods[idx, "id.type"])
> …
>
> I tried to insert the following two lines:
>        print(head(bods))
>        print(idx)
> before the last line above and rerun rerun: R CMD build --keep-empty-dirs --no-resave-data gage
> The error still exists, but I don’t see the printed out results from the newly added lines, hence not able to figure out what happened there.
> so far, I can think of anything wrong with gage package. because the manual install and vignette creating worked. The only problem I am sure is that R CMD build --keep-empty-dirs --no-resave-data behave differently from the manual process. Could you help to check into the problem here? Thank you! I am cc’ing the devel list see if other people got the same problem.
> Weijun
>
> --------------------------------------------
> On Fri, 4/4/14, Dan Tenenbaum <dtenenba at fhcrc.org> wrote:
>
>   Subject: Re: unexpected build and check error
>
>   Date: Friday, April 4, 2014, 12:24 PM
>
>   Hi Weijun,
>
>   gage is still broken today:
>
>   http://www.bioconductor.org/checkResults/2.14/bioc-LATEST/gage/zin2-buildsrc.html
>
>   It seems to be using the latest versions:
>
>   > packageVersion("gage")
>   [1] ‘2.13.5’
>   > packageVersion("pathview")
>   [1] ‘1.3.4’
>
>   So maybe the cause of the problem is different? Have you
>   tried using the most recent R? (R-3.1.0 RC)?
>
>   Dan
>
>
>
>   ----- Original Message -----
>   > From: "Dan Tenenbaum" <dtenenba at fhcrc.org>
>
>   > Sent: Thursday, April 3, 2014 10:24:10 AM
>   > Subject: Re: unexpected build and check error
>   >
>   > No, I think it will be ok tomorrow. If not I'll make
>   sure it gets
>   > fixed.
>   > Thanks,
>   > Dan
>   >
>   >
>   > ----- Original Message -----
>
>   > > To: "Dan Tenenbaum" <dtenenba at fhcrc.org>
>   > > Sent: Thursday, April 3, 2014 10:19:16 AM
>   > > Subject: unexpected build and check error
>   > >
>   > > Dear Dan,
>   > > I got the following error in the bioc 2.14 build
>   report:
>   > > http://bioconductor.org/checkResults/devel/bioc-LATEST/gage/zin2-buildsrc.html
>   > >
>   > > Error: processing vignette 'gage.Rnw' failed with
>   diagnostics:
>   > >  chunk 12 (label = go.gsets)
>   > > Error in bods[idx, "id.type"] : subscript out of
>   bounds
>   > > Execution halted
>   > >
>   > > I believe this is due to the loading of the old
>   pathview package at
>   > > that time, which carries a copy of obsolete bods
>   hence interfered
>   > > the build/check. But since pathview and gage was
>   updated at the
>   > > same
>   > > time. This error will be gone in the next
>   building/checking cycle.
>   > > My local copy does get through the
>   building/checking free of error.
>   > > Shall we do something given that the today is
>   “Deadline for
>   > > packages
>   > > passing R CMD build and check without warning”?
>   thanks!
>   > > Weijun
>   > >
>   > >
>   >
>
> _______________________________________________
> Bioc-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/bioc-devel
>


-- 
Computational Biology / Fred Hutchinson Cancer Research Center
1100 Fairview Ave. N.
PO Box 19024 Seattle, WA 98109

Location: Arnold Building M1 B861
Phone: (206) 667-2793