help in R 2.10 Windows

[Ross Boylan]

I'm using R 2.10 on Windows XP with ESS 5.5.

Since the previous discussion indicates startup scripts are problematic
for setting help, I set inferior-ess-help-command to "help(\"%s\",
help_type=\"text\")".

After this, when I use ? the system hangs until I do ctl-g.  When I then
do ctl-c ctl-c the help text prints out in the main *R* buffer.

I wanted text mostly because it took too long to get the help into the
browser (MSIE), even after the browser was opened.

Then I changed the help command to
 "help(\"%s\", help_type=\"html\")",
hoping at least to avoid the "htmlhelp is deprecated" warnings that
appeared as if they were regular ess help buffers.

However, I still have to type at least ctl-g and enter to get the help
to appear in the browser.

Any suggestions?

Thanks.
Ross Boylan

______________________________________________
ESS-help@... mailing list
https://stat.ethz.ch/mailman/listinfo/ess-help

[Ross Boylan]

On Mon, 2009-11-02 at 12:40 -0800, Ross Boylan wrote:

> I'm using R 2.10 on Windows XP with ESS 5.5.
>
> Since the previous discussion indicates startup scripts are problematic
> for setting help, I set inferior-ess-help-command to "help(\"%s\",
> help_type=\"text\")".
>
> After this, when I use ? the system hangs until I do ctl-g.  When I then
> do ctl-c ctl-c the help text prints out in the main *R* buffer.
>
> I wanted text mostly because it took too long to get the help into the
> browser (MSIE), even after the browser was opened.
>
> Then I changed the help command to
>  "help(\"%s\", help_type=\"html\")",
> hoping at least to avoid the "htmlhelp is deprecated" warnings that
> appeared as if they were regular ess help buffers.
>
> However, I still have to type at least ctl-g and enter to get the help
> to appear in the browser.
>

And then I noticed that the original strings had \n at the end.  By
adding that, the first variant works (help inside of Emacs in a separate
buffer).  If I get help on a command I already got htmlhelp for I see
the old text buffer; ordinarily that would be the right thing to do.  In
my case I see
Warning message:
In help("intervals", htmlhelp = TRUE) :
  htmhelp = TRUE is deprecated: use help_type ="html"
Killing the old buffer cleans things up.

That still leaves the problem of how to make the setting persistent so I
don't need to set the variable each session.  I'm not sure if the
previous thread on help implies I can set something in .Rprofile or not.

Ross

______________________________________________
ESS-help@... mailing list
https://stat.ethz.ch/mailman/listinfo/ess-help

[Martin Maechler-4]

Hi Ross
(and everyone else).

In the "development version" of ESS (in 'svn' / subversion)
I have committed changes that should fix this problem persistently.
I'd be glad if you could try out the new (unreleased) version of ESS
and confirm that my changes fix the problem.
You can get the Windows-zip file
  ftp://stat.ethz.ch/U/maechler/ESS/ess-5.6.zip
which you need to unpack at the appropriate place,
or possibly only use its ./lisp/ subdirectory instead of the installed
./ess/ one..

Best regards,
Martin Maechler

On Mon, Nov 2, 2009 at 22:08, Ross Boylan <ross@...> wrote:

> On Mon, 2009-11-02 at 12:40 -0800, Ross Boylan wrote:
>> I'm using R 2.10 on Windows XP with ESS 5.5.
>>
>> Since the previous discussion indicates startup scripts are problematic
>> for setting help, I set inferior-ess-help-command to "help(\"%s\",
>> help_type=\"text\")".
>>
>> After this, when I use ? the system hangs until I do ctl-g.  When I then
>> do ctl-c ctl-c the help text prints out in the main *R* buffer.
>>
>> I wanted text mostly because it took too long to get the help into the
>> browser (MSIE), even after the browser was opened.
>>
>> Then I changed the help command to
>>  "help(\"%s\", help_type=\"html\")",
>> hoping at least to avoid the "htmlhelp is deprecated" warnings that
>> appeared as if they were regular ess help buffers.
>>
>> However, I still have to type at least ctl-g and enter to get the help
>> to appear in the browser.
>>
> And then I noticed that the original strings had \n at the end.  By
> adding that, the first variant works (help inside of Emacs in a separate
> buffer).  If I get help on a command I already got htmlhelp for I see
> the old text buffer; ordinarily that would be the right thing to do.  In
> my case I see
> Warning message:
> In help("intervals", htmlhelp = TRUE) :
>  htmhelp = TRUE is deprecated: use help_type ="html"
> Killing the old buffer cleans things up.
>
> That still leaves the problem of how to make the setting persistent so I
> don't need to set the variable each session.  I'm not sure if the
> previous thread on help implies I can set something in .Rprofile or not.
>
> Ross
>
> ______________________________________________
> ESS-help@... mailing list
> https://stat.ethz.ch/mailman/listinfo/ess-help
>
>

______________________________________________
ESS-help@... mailing list
https://stat.ethz.ch/mailman/listinfo/ess-help

[Ross Boylan]

On Mon, 2009-11-02 at 23:19 +0100, Martin Maechler wrote:

> Hi Ross
> (and everyone else).
>
> In the "development version" of ESS (in 'svn' / subversion)
> I have committed changes that should fix this problem persistently.
> I'd be glad if you could try out the new (unreleased) version of ESS
> and confirm that my changes fix the problem.
> You can get the Windows-zip file
>   ftp://stat.ethz.ch/U/maechler/ESS/ess-5.6.zip
> which you need to unpack at the appropriate place,
> or possibly only use its ./lisp/ subdirectory instead of the installed
> ./ess/ one..
>

Thanks.

The out of the box behavior is what I want (help in emacs).

inferior-ess-help-command is "help(\"%s\", htmlhelp=FALSE)\n"
which is still using the obsolescent argument htmlhelp.

I'm not entirely sure I'm picking of the new code; the ESS version still
says 5.5.  I did restart emacs.

I also just (re)discovered that my .emacs file had
(setq inferior-ess-r-help-command "help(\"%s\", htmlhelp=TRUE)\n")
so my previous results need to be interpreted in that light.  Previous
discussion seemed to indicate setting the variable was irrelevant in ESS
5.5.  For tests reported in this message I commented out the setting.

Also, I reported previously a long delay getting help to pop up in a
browser.  I get the same delay on the initial launch of R inside ESS.
That supports the theory that my virus scanner is getting in the way.

Ross

> Best regards,
> Martin Maechler
>
> On Mon, Nov 2, 2009 at 22:08, Ross Boylan <ross@...> wrote:
> > On Mon, 2009-11-02 at 12:40 -0800, Ross Boylan wrote:
> >> I'm using R 2.10 on Windows XP with ESS 5.5.
> >>
> >> Since the previous discussion indicates startup scripts are problematic
> >> for setting help, I set inferior-ess-help-command to "help(\"%s\",
> >> help_type=\"text\")".
> >>
> >> After this, when I use ? the system hangs until I do ctl-g.  When I then
> >> do ctl-c ctl-c the help text prints out in the main *R* buffer.
> >>
> >> I wanted text mostly because it took too long to get the help into the
> >> browser (MSIE), even after the browser was opened.
> >>
> >> Then I changed the help command to
> >>  "help(\"%s\", help_type=\"html\")",
> >> hoping at least to avoid the "htmlhelp is deprecated" warnings that
> >> appeared as if they were regular ess help buffers.
> >>
> >> However, I still have to type at least ctl-g and enter to get the help
> >> to appear in the browser.
> >>
> > And then I noticed that the original strings had \n at the end.  By
> > adding that, the first variant works (help inside of Emacs in a separate
> > buffer).  If I get help on a command I already got htmlhelp for I see
> > the old text buffer; ordinarily that would be the right thing to do.  In
> > my case I see
> > Warning message:
> > In help("intervals", htmlhelp = TRUE) :
> >  htmhelp = TRUE is deprecated: use help_type ="html"
> > Killing the old buffer cleans things up.
> >
> > That still leaves the problem of how to make the setting persistent so I
> > don't need to set the variable each session.  I'm not sure if the
> > previous thread on help implies I can set something in .Rprofile or not.
> >
> > Ross
> >
> > ______________________________________________
> > ESS-help@... mailing list
> > https://stat.ethz.ch/mailman/listinfo/ess-help
> >
> >

______________________________________________
ESS-help@... mailing list
https://stat.ethz.ch/mailman/listinfo/ess-help

[Ross Boylan]

On Mon, 2009-11-02 at 23:19 +0100, Martin Maechler wrote:
> I'd be glad if you could try out the new (unreleased) version of ESS
> and confirm that my changes fix the problem.
> You can get the Windows-zip file
>   ftp://stat.ethz.ch/U/maechler/ESS/ess-5.6.zip
I unpacked to the top level directory before, so changes went to lisp.
However, the previous installation of ess was in site-lisp, which I
assume has precedence.  I unpacked the new lisp directory over the old
one and tried again.  Results were the same.

The variable setting is
inferior-ess-help-command is a variable defined in `ess-cust.el'.
Its value is "help(\"%s\", htmlhelp=FALSE)\n"

?foo opens an emacs buffer on foo's help.

ESS's opening message and ess-version variable both say 5.5.

Ross

______________________________________________
ESS-help@... mailing list
https://stat.ethz.ch/mailman/listinfo/ess-help

[Martin Maechler-4]

>>>>> "RossB" == Ross Boylan <ross@...>
>>>>>     on Mon, 02 Nov 2009 15:17:11 -0800 writes:

    RossB> On Mon, 2009-11-02 at 23:19 +0100, Martin Maechler
    RossB> wrote:
    >> I'd be glad if you could try out the new (unreleased)
    >> version of ESS and confirm that my changes fix the
    >> problem.  You can get the Windows-zip file
    >> ftp://stat.ethz.ch/U/maechler/ESS/ess-5.6.zip
    RossB> I unpacked to the top level directory before, so
    RossB> changes went to lisp.  However, the previous
    RossB> installation of ess was in site-lisp, which I assume
    RossB> has precedence.  I unpacked the new lisp directory
    RossB> over the old one and tried again.  Results were the
    RossB> same.

    RossB> The variable setting is inferior-ess-help-command is
    RossB> a variable defined in `ess-cust.el'.  Its value is
    RossB> "help(\"%s\", htmlhelp=FALSE)\n"

not good

    RossB> ?foo opens an emacs buffer on foo's help.

    RossB> ESS's opening message and ess-version variable both
    RossB> say 5.5.

not good.

It seems that you somehow don't pick up the new code.
Do you  by chance also have *.elc  files in the  lisp/ or
site-lisp/ directory in which you unpacked the new ess-5.6.zip

If yes, that would explain everything and you need to remove all
*.elc files in there (and then restart emacs again).

?

Regards, Martin

______________________________________________
ESS-help@... mailing list
https://stat.ethz.ch/mailman/listinfo/ess-help

[Ross Boylan]

On Tue, 2009-11-03 at 09:54 +0100, Martin Maechler wrote:
>     RossB> ESS's opening message and ess-version variable both
>     RossB> say 5.5.
>
> not good.
I think I've fixed that; ess-version and the startup message both say
5.6.

inferior-ess-help-command is a variable defined in `ess-cust.el'.
Its value is
".help.ESS(\"%s\", help_type=\"text\")\n"

However, ?nlm gets
Error: could not find function ".help.ESS"
in the nlm help buffer.

Ross

______________________________________________
ESS-help@... mailing list
https://stat.ethz.ch/mailman/listinfo/ess-help

[Martin Maechler-4]

On Tue, Nov 3, 2009 at 19:20, Ross Boylan <ross@...> wrote:

> On Tue, 2009-11-03 at 09:54 +0100, Martin Maechler wrote:
>>     RossB> ESS's opening message and ess-version variable both
>>     RossB> say 5.5.
>>
>> not good.
> I think I've fixed that; ess-version and the startup message both say
> 5.6.
>
> inferior-ess-help-command is a variable defined in `ess-cust.el'.
> Its value is
> ".help.ESS(\"%s\", help_type=\"text\")\n"
>
> However, ?nlm gets
> Error: could not find function ".help.ESS"
> in the nlm help buffer.

Hmm, .help.ESS  should be defined via R-code that is sent from ESS,
(and depends on the R version in use).
For R >= 2.10.0,   a line
 .help.ESS <- help
>
> Ross
>
>

______________________________________________
ESS-help@... mailing list
https://stat.ethz.ch/mailman/listinfo/ess-help

[Ross Boylan]

On Tue, 2009-11-03 at 23:12 +0100, Martin Maechler wrote:

> On Tue, Nov 3, 2009 at 19:20, Ross Boylan <ross@...> wrote:
> > On Tue, 2009-11-03 at 09:54 +0100, Martin Maechler wrote:
> >>     RossB> ESS's opening message and ess-version variable both
> >>     RossB> say 5.5.
> >>
> >> not good.
> > I think I've fixed that; ess-version and the startup message both say
> > 5.6.
> >
> > inferior-ess-help-command is a variable defined in `ess-cust.el'.
> > Its value is
> > ".help.ESS(\"%s\", help_type=\"text\")\n"
> >
> > However, ?nlm gets
> > Error: could not find function ".help.ESS"
> > in the nlm help buffer.
>
> Hmm, .help.ESS  should be defined via R-code that is sent from ESS,
> (and depends on the R version in use).
> For R >= 2.10.0,   a line
>  .help.ESS <- help
> >

Could the fact that ESS gets a timeout waiting for the R process to
start be relevant?
Ross

______________________________________________
ESS-help@... mailing list
https://stat.ethz.ch/mailman/listinfo/ess-help

[Martin Maechler-4]

>>>>> "RossB" == Ross Boylan <ross@...>
>>>>>     on Tue, 03 Nov 2009 15:00:30 -0800 writes:

    RossB> On Tue, 2009-11-03 at 23:12 +0100, Martin Maechler
    RossB> wrote:
    >> On Tue, Nov 3, 2009 at 19:20, Ross Boylan
    >> <ross@...> wrote: > On Tue, 2009-11-03 at
    >> 09:54 +0100, Martin Maechler wrote: >> RossB> ESS's
    >> opening message and ess-version variable both >> RossB>
    >> say 5.5.
    >> >>
    >> >> not good.  > I think I've fixed that; ess-version and
    >> the startup message both say
    >> > 5.6.
    >> >
    >> > inferior-ess-help-command is a variable defined in
    >> `ess-cust.el'.  > Its value is > ".help.ESS(\"%s\",
    >> help_type=\"text\")\n"
    >> >
    >> > However, ?nlm gets > Error: could not find function
    >> ".help.ESS" > in the nlm help buffer.
    >>
    >> Hmm, .help.ESS should be defined via R-code that is sent
    >> from ESS, (and depends on the R version in use).  For R
    >> >= 2.10.0, a line .help.ESS <- help
    >> >
    RossB> Could the fact that ESS gets a timeout waiting for
    RossB> the R process to start be relevant?  Ross

Maybe.
I just tried this on our Windows Server machine.
What I get
 with the above
 1) unpacking of ess-5.6.zip in
    C:/program files/GNU Emacs/site-lisp/ess/
 2) Deleting all *.elc  files in there  (!!)
 3) Restarting Emacs
 4) M-x R

is the following *R* buffer (indented by 1)

 ----------------------------------------------------------------

 R version 2.10.0 (2009-10-26)
 Copyright (C) 2009 The R Foundation for Statistical Computing
 ISBN 3-900051-07-0

 R is free software and comes with ABSOLUTELY NO WARRANTY.
 You are welcome to redistribute it under certain conditions.
 Type 'license()' or 'licence()' for distribution details.

 R is a collaborative project with many contributors.
 Type 'contributors()' for more information and
 'citation()' on how to cite R or R packages in publications.

 Type 'demo()' for some demos, 'help()' for on-line help, or
 'help.start()' for an HTML browser interface to help.
 Type 'q()' to quit R.

 > options(chmhelp = FALSE, help_type = "text")
 > .help.ESS <- help
 > options(STERM='iESS', editor='gnuclient.exe')

 ----------------------------------------------------------------

Note the last three lines of ESS-auto-generated R code.

Don't you get these?

Best regards,
Martin Maechler, ETH Zurich

______________________________________________
ESS-help@... mailing list
https://stat.ethz.ch/mailman/listinfo/ess-help

[Ross Boylan]

On Wed, 2009-11-04 at 10:39 +0100, Martin Maechler wrote:
>  > options(chmhelp = FALSE, help_type = "text")
>  > .help.ESS <- help
>  > options(STERM='iESS', editor='gnuclient.exe')
>
>  ----------------------------------------------------------------
>
> Note the last three lines of ESS-auto-generated R code.
>
> Don't you get these?
I do not.
Ross

______________________________________________
ESS-help@... mailing list
https://stat.ethz.ch/mailman/listinfo/ess-help

[Ross Boylan]

On Wed, 2009-11-04 at 12:16 -0800, Ross Boylan wrote:

> On Wed, 2009-11-04 at 10:39 +0100, Martin Maechler wrote:
> >  > options(chmhelp = FALSE, help_type = "text")
> >  > .help.ESS <- help
> >  > options(STERM='iESS', editor='gnuclient.exe')
> >
> >  ----------------------------------------------------------------
> >
> > Note the last three lines of ESS-auto-generated R code.
> >
> > Don't you get these?
> I do not.
> Ross
>
>

I put this in my .emacs
(setq ess-loop-timeout 200000000)
but it doesn't seem to have affected the variable after I start, which
still shows as 2000000 (I multiplied x 100).

Oh, maybe I need to put the setting in some kind of mode-specific hook?
Ross

______________________________________________
ESS-help@... mailing list
https://stat.ethz.ch/mailman/listinfo/ess-help

[Martin Maechler-4]

On Wed, Nov 4, 2009 at 22:05, Ross Boylan <ross@...> wrote:

> On Wed, 2009-11-04 at 12:16 -0800, Ross Boylan wrote:
>> On Wed, 2009-11-04 at 10:39 +0100, Martin Maechler wrote:
>> >  > options(chmhelp = FALSE, help_type = "text")
>> >  > .help.ESS <- help
>> >  > options(STERM='iESS', editor='gnuclient.exe')
>> >
>> >  ----------------------------------------------------------------
>> >
>> > Note the last three lines of ESS-auto-generated R code.
>> >
>> > Don't you get these?
>> I do not.
>> Ross
>>
>>
> I put this in my .emacs
> (setq ess-loop-timeout 200000000)
> but it doesn't seem to have affected the variable after I start, which
> still shows as 2000000 (I multiplied x 100).
>
> Oh, maybe I need to put the setting in some kind of mode-specific hook?

Yes, either that [inferior-ess-mode-hook] , or -- simpler --
(setq ess-S-loop-timeout (* 100 2000000))

Martin

______________________________________________
ESS-help@... mailing list
https://stat.ethz.ch/mailman/listinfo/ess-help

[Ross Boylan]

On Wed, 2009-11-04 at 22:15 +0100, Martin Maechler wrote:

> On Wed, Nov 4, 2009 at 22:05, Ross Boylan <ross@...> wrote:
> > On Wed, 2009-11-04 at 12:16 -0800, Ross Boylan wrote:
> >> On Wed, 2009-11-04 at 10:39 +0100, Martin Maechler wrote:
> >> >  > options(chmhelp = FALSE, help_type = "text")
> >> >  > .help.ESS <- help
> >> >  > options(STERM='iESS', editor='gnuclient.exe')
> >> >
> >> >  ----------------------------------------------------------------
> >> >
> >> > Note the last three lines of ESS-auto-generated R code.
> >> >
> >> > Don't you get these?
> >> I do not.
> >> Ross
> >>
> >>
> > I put this in my .emacs
> > (setq ess-loop-timeout 200000000)
> > but it doesn't seem to have affected the variable after I start, which
> > still shows as 2000000 (I multiplied x 100).
> >
> > Oh, maybe I need to put the setting in some kind of mode-specific hook?
>
> Yes, either that [inferior-ess-mode-hook] , or -- simpler --

That did work.  It stopped the message about timeout, and it resulted in
the expected commands being stuffed to the buffer.

I did this.  Is there a more elegant way, i.e., without the lambda?
(add-hook 'inferior-ess-mode-hook (lambda ()
   (setq ess-loop-timeout (* 100 2000000))))

emacs used all of one CPU while it was waiting for the command; maybe
some of the timeout arguments in accept-process-output in ess-inf.el
would be gentler?

Does this reset my timeout period generally, i.e. for subsequent
commands?  It's probably excessive for that.

Startup of the R subprocess is taking c. 40 seconds.  Most of that time
seems to be reading my .RData file; at 9MB it doesn't seem that large.
As alway, I suspect the virus scanner; unfortunately it is locked down
and I can't change it.

> (setq ess-S-loop-timeout (* 100 2000000))
That didn't work.
The local value (in a .R file) of ess-loop-timeout (not
ess-S-loop-timeout) was 2e6, even though the global value was 2e8.

It's not clear to me that any changes in ESS's behavior are in order.

Ross

______________________________________________
ESS-help@... mailing list
https://stat.ethz.ch/mailman/listinfo/ess-help

[Ross Boylan]

When I go to ctl-c ctl-l in a buffer with an R file, emacs says
"No ESS process is associated with this buffer now."

Select 'switch-process' through the ESS menus fixes things up.

Recipe:
start emacs (23.1) by clicking on icon.
esc-x R and select my favorite directory
wait for things to start
settings stuffed into buffer when R starts
ctl-x ctl-f to open my.R (file)
ctl-c ctl-l
error message

More typically I drag the .R file onto the emacs icon, start r (via
ess), and then try to read the file in.  Same error message.
Ross

______________________________________________
ESS-help@... mailing list
https://stat.ethz.ch/mailman/listinfo/ess-help

[Ross Boylan]

Another oddity is that rterm is using all of one CPU just sitting there
under ESS.

I don't know if this is related to my increased loop settings; the time
is being reported against rterm, not emacs.  When I launch R standalone
it (Rgui) uses only a little CPU.

Ross

______________________________________________
ESS-help@... mailing list
https://stat.ethz.ch/mailman/listinfo/ess-help

[Ross Boylan]

On Thu, 2009-11-05 at 11:06 -0800, Ross Boylan wrote:
> Another oddity is that rterm is using all of one CPU just sitting there
> under ESS.
>
> I don't know if this is related to my increased loop settings; the time
> is being reported against rterm, not emacs.  When I launch R standalone
> it (Rgui) uses only a little CPU.
>
I think the Rterm process was some kind of zombie.  Exiting (ctl-c
ctl-q) the R session under ESS made no difference in the existence or
CPU use of the Rterm.  I killed that process manually, and then
restarted R under ESS.  rterm is using no visible CPU now.

Ross

______________________________________________
ESS-help@... mailing list
https://stat.ethz.ch/mailman/listinfo/ess-help