R help - Jan 2002 - Where is the best place to get ALL the changes between versions?

There seems to be a change in the behavior of the
function ' mean() ' in 1.40. 

In earlier versions, the application of mean to a
dataframe (all numeric)would return the aggregate mean
of all of the columns. Now, in 1.40 application of
mean to the dataframe returns a list of the column
means. The new 'help' documents the new behavior. So
the change is obviously intend

I read the "changes" document distributed with 1.40
and saw no mention of 'mean'. Is there a place where
all of the changes are documented? This would be a
help in maintaining/upgrading models and would save
tedious debugging in the longer.

Also, to the development team---do something nice for
yourself today. R is really a very superb tool and JOY
to use. (LT, Lisp-stat still rates way up there too!)

Best regards,
mwgrant at adelphia.net




__________________________________________________
Do You Yahoo!?
Send FREE video emails in Yahoo! Mail!
http://promo.yahoo.com/videomail/
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

On Wed, 9 Jan 2002, Michael Grant wrote:
> There seems to be a change in the behavior of the
> function ' mean() ' in 1.40.
>
> In earlier versions, the application of mean to a
> dataframe (all numeric)would return the aggregate mean
> of all of the columns. Now, in 1.40 application of
> mean to the dataframe returns a list of the column
> means. The new 'help' documents the new behavior. So
> the change is obviously intend
>
> I read the "changes" document distributed with 1.40
> and saw no mention of 'mean'. Is there a place where
> all of the changes are documented? This would be a
> help in maintaining/upgrading models and would save
> tedious debugging in the longer.
It's called NEWS.  There is no "changes" document, but the Windows
port
has CHANGES which starts

                Windows-specific changes to R
                ============================
See the file NEWS for changes that affect all versions of R.


And NEWS says

    o   mean() has `data frame' method applying mean column-by-column.
        When applied to non-numeric data mean() now returns NA rather
        than a confusing error message (for compatibility with S4).
        Logicals are still coerced to numeric.


So it's all there if you care to look.

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272860 (secr)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

EXECUTIVE SUMMARY:"Its called the NEWS." would have
sufficed. 

FULL TEXT:

"... So it's all there if you care to look."   

Ouch!  Yes, I do care to look and I do not take
lightly, as implied by that retort the time the team
members give to make R what it is. This is especially
the case with the help. I regret if the perception is
otherwise.

Before submitting that question, I looked at four
years of r-help archives and confirmed that the
aggregate mean was calculated, i.e., the behavior in
earlier versions of my application were acting in the
manner expected at that time. I also searched the bug
reports--manually I might add, because, Oh Lord
forgive me, I've never learned regular expressions. I
found the new 'mean' behavior in the new documentation
as noted above in this thread.  Furthermore, I
confirmed the above by checking the source codes for
1.22 and 1.40 in their respective base libraries. Uh,
These things took a little of my time and effort.
That's OK. I learned from it.

All of this took place in the wee hours of the AM
after debugging(?) a MC simulation of a high level
nuclear waste inventory that all of the sudden in 1.4
went screaming to three times where it was
before--with no changes in data or code. (Not anxious
to tell the client that one, are we, huh?) 

The change in mean behavior was tracked down fairly
easily.  Of course, the real issue was: is there
anything else? 

So you see it is not necessarily the case that one is
impatient or too lazy if one submits a question that
on the surface may appear to be so. It may be that
one, like you, is up to his 'you know what' in
alligators!  That being said, I will risk another
zinger to tender some constructive advice that might
make your efforts more valuable to the R community.

One way to reduce the number of  such 'oversight'
questions is to lower the probability of their
occurring. In that regard, you MIGHT want to consider
consolidating changes into one document with platform
specific suggestions.  I  doubt I'm  the only person
who uses "search" in an editor to find something so
specific. ("Windows-specific changes to R" did have an
ice cube's chance in hell of flashing before my eyes.)

Always try to remember that things may not be as clear
to others as you, the team, perceive them to be.
Consider a line from readme.rw1040:

"See the file `CHANGES' for the NEW (my upper case) 
features of this version. "

Yet, as noted in the response to me, the changes are
under 'NEWS'. Got it. It's clear now. 


Now an observation regarding the response:

"And NEWS says 
    o mean() has `data frame' method applying mean
column-by-column. 
      When applied to non-numeric data mean() now
returns NA rather 
      than a confusing error message (for
compatibility with S4). 
      Logicals are still coerced to numeric. "

Prudence would suggest that because the use of the
'mean' is very, very common in applications, the NEWS
could provide some small caution regarding the older
behavior of mean and legacy R code. (This concept
extends to changes in other heavily used functions.)
Clearly this is an editorial comment and one might
protest that the developers' are already spending
hours of their time in coding writing and providing
exceptional help, and hence, they should not have to
waste time answering questions from those individuals
too lazy to read the documentation. I agree with this
sentiment. However, you might be able to eliminate a
lot of these questions by considering, just a little,
the consequences of changes in preparation to writing
documentation. Is time to get a volunteer technical
editor?

Finally, don't detract from the quality of the R
effort by presuming you know the attitude or intent of
a questioner. Show a little more respect for what
might appear to be a "dumb" or "lazy" question. It may
be neither. Or maybe it is, but you probably asked one
sometime in the distant past. Gentlemen, having
written and documented applications for third parties,
I truly respect the effort and time involved with
producing validated functional code for others. Also,
having successfully and unsuccessfully taught I
understand how exasperating some questions can be. But
I've never seen a lack of respect bear fruit. And
remember--you volunteered.

Respectfully,
Dr. Michael W. Grant





__________________________________________________
Do You Yahoo!?
Send FREE video emails in Yahoo! Mail!
http://promo.yahoo.com/videomail/

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._