R help - Feb 2006 - R command line: need intelligent command history recall?

Hi all,

I am not sure if this feature exists in the R-console command line prompt:

In Matlab, if I want to enter a command which is similar to what I have
entered before,

I can enter a few prefix, then press "->", the previous command
that matches
with this prefix will then appear on this command line, and it saves a lot
of our time.

For example:
> abline(lm(new~old))
> cor(new, old)
...
...
... many lines entered
...
...

now I want to reuse "abline(lm(new~old))",

R-console provides "->" functionality to recall old commands, but
it trace
back one by one, it is slow if "abline" is way back, say 50 lines
above my
current command line... it is too slow.

In Matlab, I just need to enter "ab", then press "->", if
there is no other
"ab******" between the "abline" and my current command line,
then the
console will intelligently recall "abline" back to me... Very
convinient.

Does this feature exist in R?

Any other good Integrated Developement Environment for R?

Perhaps R users are mathematicians and statisticians; but as a software
engineer myself, I found a Visual C++-like integrated developement
environment is really efficient and time-saving. It and Borland C++ Builder
basically sets standard for modern UI design for programming IDEs.

To be a good IDE, it really needs to have an embedded inline debugger. I've
asked a statistician, he said he never debugged using a break-point,
line-by-line execution debugger -- I cannot imagine this. Where is the
productivity?

I've used Tinn-R. Frankly it is quite creative. It solved the line-by-line
execution problem by copying the line and pasted it to R-console
automatically. But a lot of times clipboard generates error. And often times
the copy and paste within Tinn editor itself are problemetic. For example, I
have been never able to select a portion of a line. When I paste a
paragraph, it always erases the current line and the following lines,
instead of inserting, ...  and if I want to select one line, it always
select two lines for me... etc.

Anyway, I hope there is IDEs that are better than Tinn-R.

Thanks a lot!

	[[alternative HTML version deleted]]

Michael wrote:
> Hi all,
> 
> I am not sure if this feature exists in the R-console command line prompt:
> 
> In Matlab, if I want to enter a command which is similar to what I have
> entered before,
> 
> I can enter a few prefix, then press "->", the previous
command that matches
> with this prefix will then appear on this command line, and it saves a lot
> of our time.
> 
> For example:
> 
> 
>>abline(lm(new~old))
>>cor(new, old)
> 
> ...
> ...
> ... many lines entered
> ...
> ...
> 
> now I want to reuse "abline(lm(new~old))",
> 
> R-console provides "->" functionality to recall old commands,
but it trace
> back one by one, it is slow if "abline" is way back, say 50 lines
above my
> current command line... it is too slow.
> 
> In Matlab, I just need to enter "ab", then press
"->", if there is no other
> "ab******" between the "abline" and my current command
line, then the
> console will intelligently recall "abline" back to me... Very
convinient.
> 
> Does this feature exist in R?
> 
> Any other good Integrated Developement Environment for R?
> 
> Perhaps R users are mathematicians and statisticians; but as a software
> engineer myself, I found a Visual C++-like integrated developement
> environment is really efficient and time-saving. It and Borland C++ Builder
> basically sets standard for modern UI design for programming IDEs.
> 
> To be a good IDE, it really needs to have an embedded inline debugger.
I've
> asked a statistician, he said he never debugged using a break-point,
> line-by-line execution debugger -- I cannot imagine this. Where is the
> productivity?
> 
> I've used Tinn-R. Frankly it is quite creative. It solved the
line-by-line
> execution problem by copying the line and pasted it to R-console
> automatically. But a lot of times clipboard generates error. And often
times
> the copy and paste within Tinn editor itself are problemetic. For example,
I
> have been never able to select a portion of a line. When I paste a
> paragraph, it always erases the current line and the following lines,
> instead of inserting, ...  and if I want to select one line, it always
> select two lines for me... etc.
> 
> Anyway, I hope there is IDEs that are better than Tinn-R.
> 
> Thanks a lot!
Please, submit a bug report to Tinn-R authors, if you want to see it 
improved.
Best,

Philippe Grosjean

I guess you are using Windows, as you do not tell us.

History search facilities are available in Rterm.exe, and also in readline 
as used in the interface on Unix-alikes.  Many people use Emacs+ESS.  (In 
all cases I think the search mode is slicker than you describe for 
Matlab.) And RGui has another mechanism (read on) that is more suited to a 
GUI.

If you would like a history search in RGui.exe, please contribute the code 
to do so.  R is a volunteer project, and this was deliberately not 
implemented as no Windows user expressed an interest.

I think you will find `the productivity' is in using a higher-level 
language than C.  R does have a line-by-line debugger (called debug, so 
not hard to find).  If you want to set breakpoints etc, see package 
`debug' on CRAN.  And the `Writing R Extensions' manual in the R-devel 
version of R (to become 2.3.0) has a chapter on `Debugging'.

I teach 50 or so people to use R/S a year.  They end up with different 
patterns of working.  Some use script windows all the time, some use 
Emacs+ESS, some use a Linux command line/Rterm.exe.  I believe it is a 
mistake to think that `one size fits all'.

On Sat, 4 Feb 2006, Michael wrote:
> Hi all,
>
> I am not sure if this feature exists in the R-console command line prompt:
>
> In Matlab, if I want to enter a command which is similar to what I have
> entered before,
>
> I can enter a few prefix, then press "->", the previous
command that matches
> with this prefix will then appear on this command line, and it saves a lot
> of our time.
>
> For example:
>
>> abline(lm(new~old))
>> cor(new, old)
> ...
> ...
> ... many lines entered
> ...
> ...
>
> now I want to reuse "abline(lm(new~old))",
Try history().  This pops up a window from which you can submit one or 
more command lines (or parts of lines).
> R-console provides "->" functionality to recall old commands,
but it trace
> back one by one, it is slow if "abline" is way back, say 50 lines
above my
> current command line... it is too slow.
>
> In Matlab, I just need to enter "ab", then press
"->", if there is no other
> "ab******" between the "abline" and my current command
line, then the
> console will intelligently recall "abline" back to me... Very
convinient.
>
> Does this feature exist in R?
Yes.
> Any other good Integrated Developement Environment for R?
>
> Perhaps R users are mathematicians and statisticians; but as a software
> engineer myself, I found a Visual C++-like integrated developement
> environment is really efficient and time-saving. It and Borland C++ Builder
> basically sets standard for modern UI design for programming IDEs.
That is a matter of opinion.  If you want us to accept your opinion, you 
need to give your credentials, and you haven't even told us your name and 
affiliation.
> To be a good IDE, it really needs to have an embedded inline debugger.
I've
> asked a statistician, he said he never debugged using a break-point,
> line-by-line execution debugger -- I cannot imagine this. Where is the
> productivity?
>
> I've used Tinn-R. Frankly it is quite creative. It solved the
line-by-line
> execution problem by copying the line and pasted it to R-console
> automatically. But a lot of times clipboard generates error. And often
times
> the copy and paste within Tinn editor itself are problemetic. For example,
I
> have been never able to select a portion of a line. When I paste a
> paragraph, it always erases the current line and the following lines,
> instead of inserting, ...  and if I want to select one line, it always
> select two lines for me... etc.
>
> Anyway, I hope there is IDEs that are better than Tinn-R.
>
> Thanks a lot!
>
> 	[[alternative HTML version deleted]]
>
> ______________________________________________
> R-help at stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide!
http://www.R-project.org/posting-guide.html
>
-- 
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 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595

On 2/4/2006 3:03 AM, Michael wrote:
> Hi all,
> 
> I am not sure if this feature exists in the R-console command line prompt:
> 
> In Matlab, if I want to enter a command which is similar to what I have
> entered before,
> 
> I can enter a few prefix, then press "->", the previous
command that matches
> with this prefix will then appear on this command line, and it saves a lot
> of our time.
> 
> For example:
> 
>> abline(lm(new~old))
>> cor(new, old)
> ...
> ...
> ... many lines entered
> ...
> ...
> 
> now I want to reuse "abline(lm(new~old))",
> 
> R-console provides "->" functionality to recall old commands,
but it trace
> back one by one, it is slow if "abline" is way back, say 50 lines
above my
> current command line... it is too slow.
> 
> In Matlab, I just need to enter "ab", then press
"->", if there is no other
> "ab******" between the "abline" and my current command
line, then the
> console will intelligently recall "abline" back to me... Very
convinient.
> 
> Does this feature exist in R?
> 
> Any other good Integrated Developement Environment for R?
> 
> Perhaps R users are mathematicians and statisticians; but as a software
> engineer myself, I found a Visual C++-like integrated developement
> environment is really efficient and time-saving. It and Borland C++ Builder
> basically sets standard for modern UI design for programming IDEs.
> 
> To be a good IDE, it really needs to have an embedded inline debugger.
I've
> asked a statistician, he said he never debugged using a break-point,
> line-by-line execution debugger -- I cannot imagine this. Where is the
> productivity?
Writing such a thing is a little tricky, but should be possible if 
someone devotes enough time to it.  A couple of issues are:

  - R source code currently maintains no connection to the file it came 
from.  That would need to be added for a source level debugger.

  - Not all R functions come from source code in a file; they may have 
been entered at the console, produced as the result returned by another 
function, etc.

  - Such IDEs tend to be very platform-specific.  You can do a lot of 
work to make a nice IDE on Windows, and not be able to re-use much of it 
in other platforms.

Currently I don't know of anyone actively working on such a thing.  I 
agree with you that source-level IDEs are great for productivity, and 
I'd probably switch to one if someone else wrote it.  However, I am 
unlikely to ever have time to do the work myself.

Duncan Murdoch
> 
> I've used Tinn-R. Frankly it is quite creative. It solved the
line-by-line
> execution problem by copying the line and pasted it to R-console
> automatically. But a lot of times clipboard generates error. And often
times
> the copy and paste within Tinn editor itself are problemetic. For example,
I
> have been never able to select a portion of a line. When I paste a
> paragraph, it always erases the current line and the following lines,
> instead of inserting, ...  and if I want to select one line, it always
> select two lines for me... etc.
> 
> Anyway, I hope there is IDEs that are better than Tinn-R.
> 
> Thanks a lot!
> 
> 	[[alternative HTML version deleted]]
> 
> ______________________________________________
> R-help at stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide!
http://www.R-project.org/posting-guide.html