An embedded and charset-unspecified text was scrubbed... Name: not available Url: https://stat.ethz.ch/pipermail/bioc-devel/attachments/20070121/8f28b9c2/attachment.pl
[Bioc-devel] update packages not working
10 messages · Mark Kimpel, Seth Falcon, Martin Maechler
Hi Mark, "Kimpel, Mark William" <mkimpel at iupui.edu> writes:
I'm having difficulty auto-updating packages in R 2.5/BioC 2.0. R packages update without problem, BioC packages result in following error.
CRAN packages update because:
a) the CRAN URL is essentially hard-coded into R
b) CRAN packages, in general, have very few dependencies and those
that they have are all in the CRAN.
It seems this happened with BioC 1.9 development also. What is the acceptable protocol for updating BioC development packages?
Here's one way to upgrade all of your installed CRAN _and_
Bioconductor packages:
library("Biobase")
update.packages(repos=biocReposList())
While I'm on a tear here, has anyone ever thought of synching the version numbers for BioC and R? I don't have a problem with it, but it might help newbies keep things straight. I frequently see problems on the BioC list caused by non-synched versions of BioC and R.
I can recall some recent cases where folks have had out of date packages, but version number syncing is not going to help with that. My impression is that there have been very few, if any, recent cases of release/devel mixups. Syncing R and BioC version numbers would not, IMO, solve package install/upgrade issues and it would lead to confusion that the R and Bioconductor projects were the same (they are not).
update.packages(ask='graphics')
If the above doesn't work, send along the error messages and we'll go from there. But perhaps the bioconductor list is a better forum at that point. + seth
Seth, Regarding synching version numbers of R and BioC, I see your point about giving the impression that they are the same project, whereas in reality they are separate. It seems to me, however, that there are many cases on the list that result from a newer BioC version being used with an older R version. I think that the bioinformatics folks focus in on getting the latest and greatest packages for BioC and forget that R is also evolving. On start-up, could R run a quick check of version numbers of all packages to see if they are supported by the base R version and, if not, give a list of those that are not? It might be a more visible warning that putting a warning up when packages are loaded as these can get lost in the jumble of scrolling screen output. For myself, I simply have the commands you gave me in .Rprofile and I update every time R starts up. It makes the process a little slower, but it was the only way I could make myself remember to do it on a regular basis and it certainly beats spending time debugging something only to find that I am working with an outdated package. Well, now onto a real-world problem that I am having. I could not get your code to work for updating my BioC devel packages. It does work for CRAN. At the end of this post is a screen dump with code, error message and sessionInfo(). Any thoughts? Mark Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R.
library("Biobase")
Loading required package: tools
Welcome to Bioconductor
Vignettes contain introductory material. To view, type
'openVignette()' or start with 'help(Biobase)'. For details
on reading vignettes, see the openVignette help page.
update.packages(repos=biocReposList())
Warning: unable to access index for repository http://bioconductor.org/packages/2.0/monograph/bin/windows/contrib/2.5
sessionInfo()
R version 2.5.0 Under development (unstable) (2007-01-19 r40528) i386-pc-mingw32 locale: LC_COLLATE=English_United States.1252;LC_CTYPE=English_United States.1252;LC_MONETARY=English_United States.1252;LC_NUMERIC=C;LC_TIME=English_United States.1252 attached base packages: [1] "tools" "stats" "graphics" "grDevices" "utils" "datasets" "methods" "base" other attached packages: Biobase "1.13.34"
update.packages(ask='graphics')
--- Please select a CRAN mirror for use in this session --- Mark W. Kimpel MD (317) 490-5129 Work, & Mobile (317) 663-0513 Home (no voice mail please) 1-(317)-536-2730 FAX -----Original Message----- From: Seth Falcon [mailto:sfalcon at fhcrc.org] Sent: Monday, January 22, 2007 12:18 AM To: Kimpel, Mark William Cc: bioc-devel at stat.math.ethz.ch Subject: Re: [Bioc-devel] update packages not working Hi Mark, "Kimpel, Mark William" <mkimpel at iupui.edu> writes:
I'm having difficulty auto-updating packages in R 2.5/BioC 2.0. R packages update without problem, BioC packages result in following error.
CRAN packages update because:
a) the CRAN URL is essentially hard-coded into R
b) CRAN packages, in general, have very few dependencies and those
that they have are all in the CRAN.
It seems this happened with BioC 1.9 development also. What is the acceptable protocol for updating BioC development packages?
Here's one way to upgrade all of your installed CRAN _and_
Bioconductor packages:
library("Biobase")
update.packages(repos=biocReposList())
While I'm on a tear here, has anyone ever thought of synching the version numbers for BioC and R? I don't have a problem with it, but it might help newbies keep things straight. I frequently see problems on the BioC list caused by non-synched versions of BioC and R.
I can recall some recent cases where folks have had out of date packages, but version number syncing is not going to help with that. My impression is that there have been very few, if any, recent cases of release/devel mixups. Syncing R and BioC version numbers would not, IMO, solve package install/upgrade issues and it would lead to confusion that the R and Bioconductor projects were the same (they are not).
update.packages(ask='graphics')
If the above doesn't work, send along the error messages and we'll go from there. But perhaps the bioconductor list is a better forum at that point. + seth
"Kimpel, Mark William" <mkimpel at iupui.edu> writes:
Seth, Regarding synching version numbers of R and BioC, I see your point about giving the impression that they are the same project, whereas in reality they are separate. It seems to me, however, that there are many cases on the list that result from a newer BioC version being used with an older R version.
Please post links to posts where you think this is happening. When users follow the directions for installing Bioconductor packages using biocLite it is virtually impossible to get a mismatch. Have there been cases of such mixups? Yes. But the ones that I can recall were the result of users being confused about how to install BioC packages in the first place -- I just fail to see how version numbers would help these users.
I think that the bioinformatics folks focus in on getting the latest and greatest packages for BioC and forget that R is also evolving. On start-up, could R run a quick check of version numbers of all packages to see if they are supported by the base R version and, if not, give a list of those that are not?
If the packages were installed using any of the recommended methods, then I don't think it is possible to have a package that isn't supported by your version of R. Adding such a feature to R is something for the R email lists and not something we can do much about here at BioC HQ ;-)
For myself, I simply have the commands you gave me in .Rprofile and I update every time R starts up. It makes the process a little slower, but it was the only way I could make myself remember to do it on a regular basis and it certainly beats spending time debugging something only to find that I am working with an outdated package.
I'm glad you've found a solution that works for you. The code in your .Rprofile file keeps packages updated with respect to bug fixes, etc. -- that's fine, but it leaves me a bit confused about how this relates to arguments about R <-> BioC version number syncing or mismatches between R version and package version.
Well, now onto a real-world problem that I am having. I could not get your code to work for updating my BioC devel packages. It does work for CRAN. At the end of this post is a screen dump with code, error message and sessionInfo(). Any thoughts? Mark Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R.
library("Biobase")
Loading required package: tools
Welcome to Bioconductor
Vignettes contain introductory material. To view, type
'openVignette()' or start with 'help(Biobase)'. For details
on reading vignettes, see the openVignette help page.
update.packages(repos=biocReposList())
Warning: unable to access index for repository http://bioconductor.org/packages/2.0/monograph/bin/windows/contrib/2.5
sessionInfo()
R version 2.5.0 Under development (unstable) (2007-01-19 r40528) i386-pc-mingw32 locale: LC_COLLATE=English_United States.1252;LC_CTYPE=English_United States.1252;LC_MONETARY=English_United States.1252;LC_NUMERIC=C;LC_TIME=English_United States.1252 attached base packages: [1] "tools" "stats" "graphics" "grDevices" "utils" "datasets" "methods" "base" other attached packages: Biobase "1.13.34"
Aside from the warning message (which will be fixed, but hey, you are using devel), what didn't work?
Seth, My thoughts on version numbers were really just an aside. You are certainly correct that if new users do everything correctly they will have no problems. Here, however, is one scenario where I can see someone screwing something up. User only used R/BioC intermittently and has version R 2.1 on his machine. He reads about a new package in Bioinformatics, goes to the BioC website, downloads this new package (created to run in R 2.4.1) and installs from zip file. Couldn't this cause problems for the user? Perhaps this is just theoretical and never happens in practice. The next time I see what I think is a real world example, I'll forward it to you. Unless someone else has something to add, let's just drop this for now. My real problem is that I am not sure if the update packages code is working in R 2.5.0. Is the warning message itself a bug or is the update not working? So far I haven't been prompted to enter "y" to update a BioC package, whereas I have for CRAN packages. Thanks, Mark Mark W. Kimpel MD (317) 490-5129 Work, & Mobile (317) 663-0513 Home (no voice mail please) 1-(317)-536-2730 FAX -----Original Message----- From: Seth Falcon [mailto:sfalcon at fhcrc.org] Sent: Monday, January 22, 2007 11:52 AM To: Kimpel, Mark William Cc: Seth Falcon; bioc-devel at stat.math.ethz.ch Subject: Re: [Bioc-devel] update packages not working "Kimpel, Mark William" <mkimpel at iupui.edu> writes:
Seth, Regarding synching version numbers of R and BioC, I see your point
about
giving the impression that they are the same project, whereas in
reality
they are separate. It seems to me, however, that there are many cases
on
the list that result from a newer BioC version being used with an
older
R version.
Please post links to posts where you think this is happening. When users follow the directions for installing Bioconductor packages using biocLite it is virtually impossible to get a mismatch. Have there been cases of such mixups? Yes. But the ones that I can recall were the result of users being confused about how to install BioC packages in the first place -- I just fail to see how version numbers would help these users.
I think that the bioinformatics folks focus in on getting the latest and greatest packages for BioC and forget that R is also evolving. On start-up, could R run a quick check of version numbers of all packages to see if they are supported by the base R version and, if not, give a list of those that are not?
If the packages were installed using any of the recommended methods, then I don't think it is possible to have a package that isn't supported by your version of R. Adding such a feature to R is something for the R email lists and not something we can do much about here at BioC HQ ;-)
For myself, I simply have the commands you gave me in .Rprofile and I update every time R starts up. It makes the process a little slower, but it was the only way I could make myself remember to do it on a regular basis and it certainly beats spending time debugging something only to find that I am working with an outdated package.
I'm glad you've found a solution that works for you. The code in your .Rprofile file keeps packages updated with respect to bug fixes, etc. -- that's fine, but it leaves me a bit confused about how this relates to arguments about R <-> BioC version number syncing or mismatches between R version and package version.
Well, now onto a real-world problem that I am having. I could not get your code to work for updating my BioC devel packages. It does work
for
CRAN. At the end of this post is a screen dump with code, error
message
and sessionInfo(). Any thoughts? Mark Type 'demo()' for some demos, 'help()' for on-line help, or 'help.start()' for an HTML browser interface to help. Type 'q()' to quit R.
library("Biobase")
Loading required package: tools
Welcome to Bioconductor
Vignettes contain introductory material. To view, type
'openVignette()' or start with 'help(Biobase)'. For details
on reading vignettes, see the openVignette help page.
update.packages(repos=biocReposList())
Warning: unable to access index for repository http://bioconductor.org/packages/2.0/monograph/bin/windows/contrib/2.5
sessionInfo()
R version 2.5.0 Under development (unstable) (2007-01-19 r40528) i386-pc-mingw32 locale: LC_COLLATE=English_United States.1252;LC_CTYPE=English_United States.1252;LC_MONETARY=English_United States.1252;LC_NUMERIC=C;LC_TIME=English_United States.1252 attached base packages: [1] "tools" "stats" "graphics" "grDevices" "utils" "datasets" "methods" "base" other attached packages: Biobase "1.13.34"
Aside from the warning message (which will be fixed, but hey, you are using devel), what didn't work?
"Kimpel, Mark William" <mkimpel at iupui.edu> writes:
Unless someone else has something to add, let's just drop this for now.
Sure.
My real problem is that I am not sure if the update packages code is working in R 2.5.0. Is the warning message itself a bug or is the update not working? So far I haven't been prompted to enter "y" to update a BioC package, whereas I have for CRAN packages.
The warning message you posted should not prevent updates, but it is difficult for me to tell if something else is going on. + seth
Seth, Do you have an older, outdated WinBinary version of one of your packages for BioC 2.0? If you send it to me, I could manually install it, then restart R with my update script in place, and see if it is updated, warning or not. Would tell us IF there is a real problem, but wouldn't pinpoint the source. Still, a small step in the right direction? Mark Mark W. Kimpel MD (317) 490-5129 Work, & Mobile (317) 663-0513 Home (no voice mail please) 1-(317)-536-2730 FAX -----Original Message----- From: Seth Falcon [mailto:sfalcon at fhcrc.org] Sent: Tuesday, January 23, 2007 9:59 AM To: Kimpel, Mark William Cc: Seth Falcon; bioc-devel at stat.math.ethz.ch Subject: Re: [Bioc-devel] update packages not working "Kimpel, Mark William" <mkimpel at iupui.edu> writes:
Unless someone else has something to add, let's just drop this for now.
Sure.
My real problem is that I am not sure if the update packages code is working in R 2.5.0. Is the warning message itself a bug or is the
update
not working? So far I haven't been prompted to enter "y" to update a BioC package, whereas I have for CRAN packages.
The warning message you posted should not prevent updates, but it is difficult for me to tell if something else is going on. + seth
2 days later
Seth, Today the warning message again appeared, but shortly thereafter I was prompted to update a host of BioC packages. So, I think the warning is a bug, but not one that prevents the updates from taking place. Mark Mark W. Kimpel MD (317) 490-5129 Work, & Mobile (317) 663-0513 Home (no voice mail please) 1-(317)-536-2730 FAX -----Original Message----- From: Seth Falcon [mailto:sfalcon at fhcrc.org] Sent: Tuesday, January 23, 2007 9:59 AM To: Kimpel, Mark William Cc: Seth Falcon; bioc-devel at stat.math.ethz.ch Subject: Re: [Bioc-devel] update packages not working "Kimpel, Mark William" <mkimpel at iupui.edu> writes:
Unless someone else has something to add, let's just drop this for now.
Sure.
My real problem is that I am not sure if the update packages code is working in R 2.5.0. Is the warning message itself a bug or is the
update
not working? So far I haven't been prompted to enter "y" to update a BioC package, whereas I have for CRAN packages.
The warning message you posted should not prevent updates, but it is difficult for me to tell if something else is going on. + seth
"Kimpel, Mark William" <mkimpel at iupui.edu> writes:
Seth, Today the warning message again appeared, but shortly thereafter I was prompted to update a host of BioC packages. So, I think the warning is a bug, but not one that prevents the updates from taking place.
Thanks for the update. I will see about fixing the bug to eliminate the warning. + seth
"Seth" == Seth Falcon <sfalcon at fhcrc.org>
on Sun, 21 Jan 2007 21:18:13 -0800 writes:
Seth> Hi Mark,
Seth> "Kimpel, Mark William" <mkimpel at iupui.edu> writes:
>> I'm having difficulty auto-updating packages in R 2.5/BioC 2.0. R
>> packages update without problem, BioC packages result in following
>> error.
Seth> CRAN packages update because:
Seth> a) the CRAN URL is essentially hard-coded into R
Seth> b) CRAN packages, in general, have very few dependencies and those
Seth> that they have are all in the CRAN.
not quite (but almost, I agree),
e.g., there is at least one CRAN package that depends on BioC 'graph'.
[................]
Martin