R help - Jul 2019 - sd in aggregate and help.search/?? seem not to work as usual in latest version

Hello dear helpers,

I am currently running a small R-crash course for beginners at my
university, since I believe that there a far too few lectures about how to
use R.

Thereby, I showed the '??' or 'help.search()' function to the
students as
well as the 'aggregate()' function with 'FUN=sd'.

 

The code worked fine for me - as usual (in version 3.5.1) - but not for the
students. Doing some research got me to the point, that in the current
version (3.6.0),

'aggregate' cannot use 'sd' for data.frames with mixed columns
of factors
and numeric columns and throws an error (which is somewhat ok, since
'var()'
is not applicable for factors, of course)  instead of putting 'NA's into
the
respective columns like it does with 'FUN=mean' and like it does in the
former version(s). 

Is this a bug or a feature? It would be nice to work with it, the way I am
used to. Of course it is not difficult to work around this problem, but it
is somewhat strange, why it was changed that way (if this was intended).

 

Additionally, the 'help.search()' always just gives the result: "No
results
found", for all of my students.

 

I didn't find any post about that issues so far, so I am not sure whether
this are known issues (but all of my 10 students had this problem).

 

If this were intended changes, I would be interested, what the reason is and
whether there are direct workarounds (e.g. instead of using only numeric
columns), for those issues. 

Otherwise I am looking forward to a fixed version. 

 

Thanks in advance for your help!

 

Kind regards,

Marvin Kiene

 

Doctoral Student

Department of Animal Ecology I, AG Prof. Laforsch

University of Bayreuth

95440 Bayreuth

Germany

 


	[[alternative HTML version deleted]]

Hi Marvin,
One way to get around the problem with "sd" is to only process those
columns of a data frame for which the variance is defined. It is an
opportunity to show students how  to write wrapper functions as well:

sd_num<-function(x) return(ifelse(is.numeric(x),sd(x),NA))
df<-data.frame(group=rep(0:1,each=5),value=1:10,name=LETTERS[1:10])
aggregate(df[,2:3],by=df[1],FUN="sd_num")

Jim

On Tue, Jul 2, 2019 at 3:47 AM Marvin Kiene <maki at maki-science.org>
wrote:
>
> Hello dear helpers,
>
> I am currently running a small R-crash course for beginners at my
> university, since I believe that there a far too few lectures about how to
> use R.
>
> Thereby, I showed the '??' or 'help.search()' function to
the students as
> well as the 'aggregate()' function with 'FUN=sd'.
>
>
>
> The code worked fine for me - as usual (in version 3.5.1) - but not for the
> students. Doing some research got me to the point, that in the current
> version (3.6.0),
>
> 'aggregate' cannot use 'sd' for data.frames with mixed
columns of factors
> and numeric columns and throws an error (which is somewhat ok, since
'var()'
> is not applicable for factors, of course)  instead of putting 'NA's
into the
> respective columns like it does with 'FUN=mean' and like it does in
the
> former version(s).
>
> Is this a bug or a feature? It would be nice to work with it, the way I am
> used to. Of course it is not difficult to work around this problem, but it
> is somewhat strange, why it was changed that way (if this was intended).
>
>
>
> Additionally, the 'help.search()' always just gives the result:
"No results
> found", for all of my students.
>
>
>
> I didn't find any post about that issues so far, so I am not sure
whether
> this are known issues (but all of my 10 students had this problem).
>
>
>
> If this were intended changes, I would be interested, what the reason is
and
> whether there are direct workarounds (e.g. instead of using only numeric
> columns), for those issues.
>
> Otherwise I am looking forward to a fixed version.
>
>
>
> Thanks in advance for your help!
>
>
>
> Kind regards,
>
> Marvin Kiene
>
>
>
> Doctoral Student
>
> Department of Animal Ecology I, AG Prof. Laforsch
>
> University of Bayreuth
>
> 95440 Bayreuth
>
> Germany
>
>
>
>
>         [[alternative HTML version deleted]]
>
> ______________________________________________
> R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.

On 1/07/19 9:41 PM, Marvin Kiene wrote:
> Hello dear helpers,
> 
> I am currently running a small R-crash course for beginners at my
> university, since I believe that there a far too few lectures about how to
> use R.
> 
> Thereby, I showed the '??' or 'help.search()' function to
the students as
> well as the 'aggregate()' function with 'FUN=sd'.
> 
>   
> 
> The code worked fine for me - as usual (in version 3.5.1) - but not for the
> students. Doing some research got me to the point, that in the current
> version (3.6.0),
> 
> 'aggregate' cannot use 'sd' for data.frames with mixed
columns of factors
> and numeric columns and throws an error (which is somewhat ok, since
'var()'
> is not applicable for factors, of course)  instead of putting 'NA's
into the
> respective columns like it does with 'FUN=mean' and like it does in
the
> former version(s).
> 
> Is this a bug or a feature? It would be nice to work with it, the way I am
> used to. Of course it is not difficult to work around this problem, but it
> is somewhat strange, why it was changed that way (if this was intended).
> 
>   
> 
> Additionally, the 'help.search()' always just gives the result:
"No results
> found", for all of my students.
> 
>   
> 
> I didn't find any post about that issues so far, so I am not sure
whether
> this are known issues (but all of my 10 students had this problem).
> 
>   
> 
> If this were intended changes, I would be interested, what the reason is
and
> whether there are direct workarounds (e.g. instead of using only numeric
> columns), for those issues.
> 
> Otherwise I am looking forward to a fixed version.
> 
>   
> 
> Thanks in advance for your help!
Can't offer much help, especially in respect of the first issue 
involving aggregate(). (But it sounds to me more like a bug than a feature.)

I would just like to point out that when I do

     help.search()

(with no argument) I get:
> Error in help.search() : do not know what to search
This seems reasonable, apart from being ungrammatical.

If I do, e.g.,

    help.search("ellipse")

I get a brazillion results.

This is with version 3.6.0.

cheers,

Rolf Turner

-- 
Honorary Research Fellow
Department of Statistics
University of Auckland
Phone: +64-9-373-7599 ext. 88276

On 01/07/2019 5:41 a.m., Marvin Kiene wrote:
> Additionally, the 'help.search()' always just gives the result:
"No results
> found", for all of my students.
Those students are probably using RStudio.  This appears to be an 
incompatibility in the RStudio browser to a nearly unannounced (as far 
as I can see) change in the help system.  You can read about the change 
in the comments around line 700 of

https://svn.r-project.org/R/trunk/src/library/utils/R/help.search.R

Perhaps if anyone from RStudio is reading this, they can read those 
comments and fix things in their front end.

To work around it, students can set

options(browser="/path/to/their/browser")

where that path will depend on what system they're using.  On MacOS, 
options(browser="/usr/bin/open") works.  On Windows, 
options(browser=NULL) works.  I don't know the R default on Linux.

This will cause lots of things to appear in an external browser instead 
of in the RStudio display, so it's not the nicest workaround, but at 
least it's fairly easy.

Duncan Murdoch

On 01/07/2019 8:28 p.m., Duncan Murdoch wrote:
> On 01/07/2019 5:41 a.m., Marvin Kiene wrote:
> 
>> Additionally, the 'help.search()' always just gives the result:
"No results
>> found", for all of my students.
> 
> Those students are probably using RStudio.  This appears to be an
> incompatibility in the RStudio browser to a nearly unannounced (as far
> as I can see) change in the help system.  You can read about the change
> in the comments around line 700 of
> 
> https://svn.r-project.org/R/trunk/src/library/utils/R/help.search.R
> 
> Perhaps if anyone from RStudio is reading this, they can read those
> comments and fix things in their front end.
> 
> To work around it, students can set
> 
> options(browser="/path/to/their/browser")
> 
> where that path will depend on what system they're using.  On MacOS,
> options(browser="/usr/bin/open") works.  On Windows,
> options(browser=NULL) works.  I don't know the R default on Linux.
> 
> This will cause lots of things to appear in an external browser instead
> of in the RStudio display, so it's not the nicest workaround, but at
> least it's fairly easy.
I just saw here:

https://github.com/rstudio/rstudio/issues/4819

that the issue is known and fixed in RStudio, but still only in a 
preview release available here:

https://www.rstudio.com/products/rstudio/download/preview/

Duncan Murdoch