Menu

#156 get_description_element finds empty lists in df descriptions

1.0
open
None
2016-05-17
2016-05-17
No

use case:
from the GUI load ia, ldt, run Andrews_92_short
then from Results do a merge.R
the y_key does not get set.
The reason is that the get_description_element cmd for filling levels is expecting that the "value" field of (Andrews_92_short colnames) contains the col names. But that's empty:
xml (Andrews_92_short colnames)
...
ANSWER: <list label="value">
ANSWER: </list>

Discussion

  • Michele Gubian

    Michele Gubian - 2016-05-17

    I see that in merge.R the argument:
    y_key=factor(eN_dollar_arg(df),levels=colnames(y))
    makes the merge actually work, but as far as I see this works only if the keys have identical names, and in any case the levels do not appear in the merge form, which is confusing.

     
  • James S. Adelman

    I believe r1305 avoids sending get_description_element when this will be ineffective.

     
  • Michele Gubian

    Michele Gubian - 2016-05-17

    I see that, it works, thanks!
    A minor thing, in merge.R this eN_dollar_arg(df) does appear in the y_key field and it gives a GUI error. This is a bit annoying for the user I guess. Any way to avoid that?

     
    • James S. Adelman

      My intent is that lazyNut will get R to parse eN_dollar_arg(df), so it will be invisible to the GUI (though the dependency on df will be signalled). But getting the value of df into this parse is a delicate operation that I haven't done yet.

       

Log in to post a comment.

MongoDB Logo MongoDB