h2oai / deepwater

Deep Learning in H2O using Native GPU Backends
Apache License 2.0
282 stars 93 forks source link

R installation error #34

Closed parunach closed 7 years ago

parunach commented 7 years ago

When I to install deep water R using

sudo R CMD INSTALL h2o_3.10.3.99999.tar.gz

I get an error

This seems to be a description error. Any thoughts?

ganeshkrishnan1 commented 7 years ago

Just start R and use the internal command to install the R packages. "install.packages"

nessaile commented 7 years ago

I get the same error, and @ganeshkrishnan1 's suggestion does not resolve the issue.

pascaloettli commented 7 years ago

Same error error here. The problem is due to the absence of xgboost.R in the section Collate inside the DESCRIPTION. You can add xgboost.R to the list in Collate section. Then run:

R CMD check h2o
R CMD build h2o
R CMD INSTALL h2o_3.10.3.99999.tar.gz
vhp1360 commented 7 years ago

Hi, I faced the same error, how can I solve it?

Thanks,

arnocandel commented 7 years ago

Sorry about that - some build cleanup issue.

Updated links to pre-built packages for R/Py/Standalone, see README.md

vhp1360 commented 7 years ago

thanks, but the same error result.

arnocandel commented 7 years ago

https://files.slack.com/files-pri/T0329MHH6-F57GWRZBL/h2o_3.11.0.99999.tar.gz

vhp1360 commented 7 years ago

I redownload it and try R CMD INSTALL h2o_3.11.0.99999.tar.gz , got

Error in getOctD(x, offset, len) : invalid octal digit

error

mdymczyk commented 7 years ago

@vhp1360 that usually happens when the download of the tar file was unsuccessful, can you untar it or does it blow up with errors? (tar xvzf h2o_3.11.0.99999.tar.gz).

vhp1360 commented 7 years ago

@mdymczyk thanks, I used @arnocandel link and got that error, today I tried the main Link and it works.

pascaloettli commented 7 years ago

Fortunately, I gave the correct way to install the package last month...

mdymczyk commented 7 years ago

@vhp1360 cool!

@pascaloettli yes, though with the latest version the xgboost entry should not be necessary. Thanks for the input, though!