[Teammetrics-discuss] Graph generation of liststat.py

[Andreas Tille]

Hi,

On Mon, Feb 13, 2012 at 12:09:42PM +0530, Sukhbir Singh wrote:
>     /home/sukhbir/teammetrics/out.tar.bz2 @ blends

I just fetched the plain *.png *and* *.txt files (because the *.txt
files are needed to track down the problem because these actually are
feeded into the R script).
 
> Have a look at them, they look good, except the old error:
> 
>     Error in `rownames<-`(`*tmp*`, value = 2012) :
>       attempt to set rownames on object with no dimensions
>     Execution halted
>     Error in `rownames<-`(`*tmp*`, value = c(2007, 2008)) :
>       attempt to set rownames on object with no dimensions
>     Execution halted
> 
> We seem to be missing many graphs because of this, like for example,
> teammetrics-discuss is missing.

$ cat commitstat_debianjr_year.txt 
 year   Ben_A 
 2007   24 
 2008   2 

$ cat authorstat_teammetrics-discuss_year.txt 
 year   Sukhbir_S       Andreas_T       Alexander_W     Scott_H         Christian_P     Tollef_Fog_H    Cord_B  Martin_Z        Stephen_G       David_B         Ana_G   Charles_P 
 2011   386     273     15      5       3       2       2       1       1       1       1       1 

$ cat commitstat_pkg-python_year.txt 
 year   doko 
 2003   1 

> Just bringing it to your attention so you can fix it when you are free.

This somehow fits my suspicion that the data are somehow broken.  These
have either only a single row or column (or both in the latest example). 
My suggestion to fix this issue would be:

  1. Drop debianjr and pkg-python from commitstats because
     we are investigating *teams*
  2. Check why teammetrics-discuss does not contain data for 2012
     I remember clearly that I have watched a proper graph of this
     list - just checked: On 2012-02-03 I have created a graph of
     teammetrics-discuss which contained also data from 2012.

Otherwise a proper "solution" for the "R problem" would be to verify
using wc to guaratee more than 2 lines of data in the txt file and some
other "column counting tool" to guarantee more than 2 columns before
feeding the data into R.  Most probably this could also be done in R
but I'm also no expert here.

Kind regards

       Andreas.

-- 
http://fam-tille.de