[datatable-help] data.table 1.4

[Matthew Dowle]

Thanks for looking Tom - the extra eyes much appreciated.

Looked into it some more. Heres a full run down ...

All 4 build logs (lin32, lin64, win and mac) state svn rev 61 at the top
(i.e. not up to date). Heres the linux64 one :
http://r-forge.r-project.org/R/?group_id=240&log=build_src&pkg=data.table&flavor=patched

For some reason, just on lin64, the check logs (but not the build logs -
that is still at 61) *have* picked up rev 80.  [None of the check logs
for the other systems (lin32, win32 or mac) have log files available.]

Interesting thing is that check log for *patched* (R 2.10.1) is OK on
lin64. This has a very recent (i.e. late) timestamp of "Sat Apr 17
16:39:34 2010", about the same time you replied. I've just noticed it.
http://r-forge.r-project.org/R/?group_id=240&log=check_x86_64_linux&pkg=data.table&flavor=patched

so its just devel (R 2.12.0) which has the error in test.data.table
(same log you posted) on lin64 :
http://r-forge.r-project.org/R/?group_id=240&log=check_x86_64_linux&pkg=data.table&flavor=devel 

That could be a bug in R 2.12.0-devel, rather than data.table. Or it
could be a bug in data.table that becomes apparent in R 2.12, or perhaps
due to a change in 2.12.   Good to have advance notice.

But the rest seems to be something up with R-forge.

Because R-forge has been growing in use so much (more packages, plus
history growing in each package I guess) the exportation at 20:30 CET is
taking longer and longer. The build machines start their rsync at a set
time after that,  so if the rysnc starts before the exportation has
finished, they miss the update.  Thats what Stefan told me last time
this happened a month or so ago, something like that anyway.  I guess
its tricky to coordinate the various machines and operating systems.
They are about to release a new system to solve these problems which
involves an intra-day build cycle.  In the meantime, we should (if I'm
understanding correctly) see it all update tomorrow.... leaving us with
a possible R 2.12.0 issue to investigate and report to r-devel (if its
not already been detected).

Heres the ticket I raised to r-forge, if exra eyes help :

https://r-forge.r-project.org/tracker/?func=detail&aid=898&group_id=34&atid=194

I hadn't looked at the lin64 logs then, as I tend to just look at the
lin32 one of the two linux columns.


On Sat, 2010-04-17 at 12:20 -0400, Short, Tom wrote:
> Matthew, it seemed to error in test.data.table according to this link:
> 
>  
> 
> http://r-forge.r-project.org/R/?group_id=240&log=check_x86_64_linux&pkg=data.table&flavor=devel 
> 
>  
> 
> - Tom
> 
>  
> 
> -----Original Message-----
> From: datatable-help-bounces at lists.r-forge.r-project.org
> [mailto:datatable-help-bounces at lists.r-forge.r-project.org] On Behalf
> Of Matthew Dowle
> Sent: Saturday, April 17, 2010 10:09 AM
> To: datatable-help at lists.r-forge.r-project.org
> Subject: [datatable-help] data.table 1.4
> 
>  
> 
> 1.4 is merged to mainline now.
> 
>  
> 
> Main change is further speed improvements for grouping e.g. 10 times
> 
> faster than tapply and aggregate.
> 
>  
> 
> Although the r-forge package tab displays its at 1.4, it seems it
> didn't
> 
> fully update behind the scenes last night, have raised ticket.
> Similar
> 
> things have happened before, and if its the same, it should hopefully
> 
> update tonight. In the meantime, install.packages() will probably
> still
> 
> grab v1.3.  It might even install v1.3 but report that its installed
> 
> v1.4.  Will check again tomorrow.
> 
>  
> 
> Matthew
> 
>  
> 
>  
> 
>  
> 
> _______________________________________________
> 
> datatable-help mailing list
> 
> datatable-help at lists.r-forge.r-project.org
> 
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/datatable-help
> 
> 
> _______________________________________________
> datatable-help mailing list
> datatable-help at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/datatable-help