data.table 1.8.2 now on CRAN

[Matthew Dowle]

All,

data.table 1.8.1 was on R-Forge for a while with quite a few people
upgrading to it. Thank you for all the reports and feedback. All known
bugs have been cleared and released as 1.8.2 to CRAN. Even if you very
recently upgraded to 1.8.1, please upgrade to 1.8.2 to receive last minute
bug fixes.

The NEWS is lengthy but please do go through it with fine toothcomb :

https://r-forge.r-project.org/scm/viewvc.php/pkg/NEWS?view=markup&root=datatable

CRAN checks are passing on Linux and Mac. Such is the maturity, automation
and strain on CRAN that an estimate for the Windows binary build exists:
for R release at 02:40am CEST 19 July, and oldrelease at 16:40.

I am on holiday next week for 2 weeks. But if you have any problems you
can always roll back to 1.8.0.

Matthew

http://datatable.r-forge.r-project.org/



_______________________________________________
datatable-help mailing list
[hidden email]
https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/datatable-help

[Matthew Dowle]

Due to some off list comments and questions :

1. Although on CRAN, it may not have found its way to all mirrors yet. If
you can't wait, check the CRAN mirror status page :

    http://cran.r-project.org/mirmon_report.html

For example if you're in Australia that currently shows cran.csiro.au as
24 hours stale (iiuc) and which mirrors are more up to date.

2. Don't (now) install from R-Forge. It's been in "building" state for
over a day now due to upstream issues (I guess). Can't trust the tarball
either from R-Forge (i.e. installing from source from R-Forge), stale. The
*only* method that is always safe on R-Forge is an 'svn up' followed by R
CMD build followed by R CMD INSTALL.

3. We always get one or two bug reports just after release from
installations where, somehow, it turns out the upgrade hasn't properly
installed. This appears to be a (likely rare) R thing which I've never
worked out. I suspect it's related to having older R sessions (possibly
zombie) holding some sort of lock. If this is raised as a possibility,
users have to date always reported that a reboot and re-install of
data.table fixed it. I've never experienced this myself. But I'm careful
to close down all R sessions, check with "ps aux | grep R", first, before
upgrading. If you observe a failed upgrade and work out why, please let me
know!

4. If you're on Windows, the Windows binary hasn't been built on CRAN yet
but the estimate is tomorrow. Only Linux and Mac versions are available
from CRAN already.

Anyway ... to check your installation that's why test.data.table() is
provided for user use, and, why in the NOTES section of NEWS it is written
how many tests that version has.

To test your installation :

require(data.table)
test.data.table()
... snip ...
All 717 tests in test.data.table() completed ok in 54.099sec

If it says 717, then you're 100% good.

Matthew

>
> All,
>
> data.table 1.8.1 was on R-Forge for a while with quite a few people
> upgrading to it. Thank you for all the reports and feedback. All known
> bugs have been cleared and released as 1.8.2 to CRAN. Even if you very
> recently upgraded to 1.8.1, please upgrade to 1.8.2 to receive last minute
> bug fixes.
>
> The NEWS is lengthy but please do go through it with fine toothcomb :
>
> https://r-forge.r-project.org/scm/viewvc.php/pkg/NEWS?view=markup&root=datatable
>
> CRAN checks are passing on Linux and Mac. Such is the maturity, automation
> and strain on CRAN that an estimate for the Windows binary build exists:
> for R release at 02:40am CEST 19 July, and oldrelease at 16:40.
>
> I am on holiday next week for 2 weeks. But if you have any problems you
> can always roll back to 1.8.0.
>
> Matthew
>
> http://datatable.r-forge.r-project.org/
>
>
>
> _______________________________________________
> datatable-help mailing list
> [hidden email]
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/datatable-help
>

_______________________________________________
datatable-help mailing list
[hidden email]
https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/datatable-help

[Karl Ove Hufthammer]

Matthew Dowle wrote:

> To test your installation :
>
> require(data.table)
> test.data.table()
> ... snip ...
> All 717 tests in test.data.table() completed ok in 54.099sec
>
> If it says 717, then you're 100% good.

Note that test.data.table() won’t necessarily work properly under a
non-English locale, as some of the warnings or error messages may
come out translated. On my (Norwegian Nynorsk) Windows system, 31 tests
failed. But if I start R from the command line by typing

  SET LC_ALL=C
  c:\Path-To-R...\Rgui.exe

everything worked fine.

--
Karl Ove Hufthammer

_______________________________________________
datatable-help mailing list
[hidden email]
https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/datatable-help

[Matthew Dowle]

> Matthew Dowle wrote:
>
>> To test your installation :
>>
>> require(data.table)
>> test.data.table()
>> ... snip ...
>> All 717 tests in test.data.table() completed ok in 54.099sec
>>
>> If it says 717, then you're 100% good.
>
> Note that test.data.table() won’t necessarily work properly under a
> non-English locale, as some of the warnings or error messages may
> come out translated. On my (Norwegian Nynorsk) Windows system, 31 tests
> failed. But if I start R from the command line by typing
>
>   SET LC_ALL=C
>   c:\Path-To-R...\Rgui.exe
>
> everything worked fine.

Never knew that before. Thanks for letting us know.

Maybe test.data.table() can Sys.setlocale() to an English one for the
duration of the tests, then, and then set it back to what it was. Now
raised as FR#2156 :

https://r-forge.r-project.org/tracker/index.php?func=detail&aid=2156&group_id=240&atid=978

I raised something similar on r-devel about the new non ascii bug fix
tests (thanks to your bug report) not working on CRAN's Mac due to its C
locale, but didn't here back yet :

https://stat.ethz.ch/pipermail/r-devel/2012-July/064419.html

Those tests of non ascii characters (708-712) had to be removed (fully
removed from the test file so it can be read and parsed, not just
commented out) until I'm advised how that should be done.  CRAN's Mac is
the only problem there. Those non-ascii tests work on all other CRAN
machines and even on R-Forge's Mac.

Matthew

> --
> Karl Ove Hufthammer
>
> _______________________________________________
> datatable-help mailing list
> [hidden email]
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/datatable-help


_______________________________________________
datatable-help mailing list
[hidden email]
https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/datatable-help