Dear R users,
in the last weeks I've looked for a better R integration into TM.
Finally I came up with these three bundles:
1) R In principal I rewrote all commands to be more flexible and to avoid to call R to get some information (except for the command 'Tidy'). I.e to show e.g. a function signature/help/insert a function parameter/completion etc. the command will look for it within R's help file system. This is not only faster but also one has now access to all installed packages. In addition to that these commands also parse the entire R script for function declarations. Please note if a command using that index will be invoked for the first time or if you deleted or installed a package the index file will be rebuilt, and this takes some milliseconds.
I also fine-tuned Rmate. Now all plot are displayed inside of the output window as thumbnails (click on it to open; drag it to save it as pdf). Furthermore now it is also possible to send a very large R script to Rmate. By using the old approach Rmate's pipe overflowed.
2) R Console (Rdaemon) I fine-tuned the entire Rdaemon. I used it many days and it turned out it's stable, fast, and you're more flexible once you understood the underlying philosophy. In other words it makes really fun. Of course, R is rather complex and I'm sure that I missed some things.
3) R Console (R.app) In that bundle you will find all commands to remote the standard R.app.
Both Console bundles are extensions to the R bundle, meaning they make usage of commands, snippets, etc. defined in the R bundle.
For each bundle I tried to write an help file. In beforehand sorry for my terrible English, it's fast written and not yet proof-read.
The best thing to start is to read the help file and have a look at the bundle menus.
!!VERY IMPORTANT!! Before installing these bundles please backup the 'old' R bundle (maybe, what I did is a flop ;) and delete the bundles R and - if it exists - Rdaemon; quit TM; start TM; and then install the new bundles.
All these bundles have beta status. Thus I would be very appreciated if someone could test them and give me any feedback regarding: -what doesn't work -what is illogical or unclear -bad arrangement or key binding -what's missing/wrong/unclear in the help file ...
plus what Mac OSX version, ppc or intel, which R version
If Allan is back I will upload the new R bundle to the repository and both Console bundles to the test repository.
Cheers,
Hans
Let me take this email as the opportunity to let everybody know that Hans has graciously agreed to take over the maintenance of the R bundle and its friends, so you are actually going to see a lot of updates to these bundles in the near future than you saw in the past. Once Allan is back and Hans gains repository access, these bundles will hopefully start the process of getting themselves into the repository. In any case, please direct questions about the R bundle and friends to Hans.
This is further a good opportunity to ask people about the R Console (R.app) bundle. I personally don't find it very useful when R daemon is around, and would propose phasing it out. We didn't want to take such a drastic step without asking the users, so please let us know if you are still using the old R Console instead of R daemon, and if so what the reasons for not switching to R daemon are.
Hans, thanks again for taking over.
Haris Skiadas Department of Mathematics and Computer Science Hanover College
On Jan 30, 2008, at 5:42 AM, Hans-Joerg Bibiko wrote:
Dear R users,
in the last weeks I've looked for a better R integration into TM.
Finally I came up with these three bundles:
- R
In principal I rewrote all commands to be more flexible and to avoid to call R to get some information (except for the command 'Tidy'). I.e to show e.g. a function signature/help/insert a function parameter/completion etc. the command will look for it within R's help file system. This is not only faster but also one has now access to all installed packages. In addition to that these commands also parse the entire R script for function declarations. Please note if a command using that index will be invoked for the first time or if you deleted or installed a package the index file will be rebuilt, and this takes some milliseconds.
I also fine-tuned Rmate. Now all plot are displayed inside of the output window as thumbnails (click on it to open; drag it to save it as pdf). Furthermore now it is also possible to send a very large R script to Rmate. By using the old approach Rmate's pipe overflowed.
- R Console (Rdaemon)
I fine-tuned the entire Rdaemon. I used it many days and it turned out it's stable, fast, and you're more flexible once you understood the underlying philosophy. In other words it makes really fun. Of course, R is rather complex and I'm sure that I missed some things.
- R Console (R.app)
In that bundle you will find all commands to remote the standard R.app.
Both Console bundles are extensions to the R bundle, meaning they make usage of commands, snippets, etc. defined in the R bundle.
For each bundle I tried to write an help file. In beforehand sorry for my terrible English, it's fast written and not yet proof-read.
The best thing to start is to read the help file and have a look at the bundle menus.
!!VERY IMPORTANT!! Before installing these bundles please backup the 'old' R bundle (maybe, what I did is a flop ;) and delete the bundles R and - if it exists - Rdaemon; quit TM; start TM; and then install the new bundles.
All these bundles have beta status. Thus I would be very appreciated if someone could test them and give me any feedback regarding: -what doesn't work -what is illogical or unclear -bad arrangement or key binding -what's missing/wrong/unclear in the help file ...
plus what Mac OSX version, ppc or intel, which R version
If Allan is back I will upload the new R bundle to the repository and both Console bundles to the test repository.
Cheers,
Hans
<R.tmbundle.zip> <R Console (Rdaemon).tmbundle.zip> <R Console (R_app).tmbundle.zip>
Hi,
thanks for the first comments!
One main problem occurred with Leopard X11.
Leopard ships with the new X11R7 server and its behavior differs from Tiger. First the good news, X11 starts automatically if an application needs it. Thus I removed the code for starting X11 automatically. (For Tiger you have to do it manually) Then next issue is Tiger X11 needs the DISPLAY variable, Leopard not – or better if it's set Leopard X11 crashes. Thus I have to distinguish between Tiger and Leopard in the source code ;(
Anyway, in principal it works, but Leopard X11 is very buggy yet and in some cases very slow (try to run fix() with a matrix - the tk window will be refreshed and refreshed and refreshed and ...).
Fine. The next item is that Rmate on Leopard (Intel) running R devel gives an error about popen3, i.e it does not run. (?) Then only thing I can say is that Rmate runs on Leopard (ppc) with the latest stable R version 2.6.1. Maybe someone could check this with an intel mac. (I read in the internet that some tiny Ruby code snippets wasn't compiled correctly for Leopard. And, R devel on Leopard should be taken with a pinch of salt yet)
OK. Please go ahead to test the R stuff. (BTW next week I won't be able to check my mails)
Cheers,
Hans
On Feb 1, 2008, at 3:23 AM, Hans-Jörg Bibiko wrote:
OK. Please go ahead to test the R stuff. (BTW next week I won't be able to check my mails)
Is the plan to have three bundles or to figure out if people don't use the old type and merge two? Always best to combine if possible. :)
On 1 Feb 2008, at 12:37, Michael Sheets wrote:
On Feb 1, 2008, at 3:23 AM, Hans-Jörg Bibiko wrote:
OK. Please go ahead to test the R stuff. (BTW next week I won't be able to check my mails)
Is the plan to have three bundles or to figure out if people don't use the old type and merge two? Always best to combine if possible. :)
If I understood you correctly, you are only talking about the R.bundle. The normal R.bundle is standard for writing R code and the new one will replace the old R.bundle.
Both other Console Bundles - by my opinion - are more or less extensions to the R.bundle.
The question is whether there is a need to have the remote bundle for R.app. Well, by myself, I'm only using the Rdaemon, but I don't know if there are users who wants to have this R.app Console bundle. (And of course, I don't know whether Rdaemon could manage ALL R tasks)
In order to avoid to mess up the user's bundle menu I subdivided the R stuff into three parts. By doing so each user can install that Console bundle which s/he wants to use.
Cheers,
Hans
Hans,
The new package is great. Rmate in particular is greatly improved!
As for the R.app console, I didn't use it very much and find the daemon much nicer to work with.
Using this version of the bundle when I use "send selection to R.app" I get the following error:
115:116: syntax error: Expected end of line, etc. but found """. (-2741)
The version you uploaded a few days ago, worked fine.
On another note, I thought the shortcut for that command used to be shift-command-R. It's hard to remember this changed and I find the new one very difficult to type.
Thanks
Peter
On Feb 1, 2008 4:14 AM, Hans-Joerg Bibiko bibiko@eva.mpg.de wrote:
On 1 Feb 2008, at 12:37, Michael Sheets wrote:
On Feb 1, 2008, at 3:23 AM, Hans-Jörg Bibiko wrote:
OK. Please go ahead to test the R stuff. (BTW next week I won't be able to check my mails)
Is the plan to have three bundles or to figure out if people don't use the old type and merge two? Always best to combine if possible. :)
If I understood you correctly, you are only talking about the R.bundle. The normal R.bundle is standard for writing R code and the new one will replace the old R.bundle.
Both other Console Bundles - by my opinion - are more or less extensions to the R.bundle.
The question is whether there is a need to have the remote bundle for R.app. Well, by myself, I'm only using the Rdaemon, but I don't know if there are users who wants to have this R.app Console bundle. (And of course, I don't know whether Rdaemon could manage ALL R tasks)
In order to avoid to mess up the user's bundle menu I subdivided the R stuff into three parts. By doing so each user can install that Console bundle which s/he wants to use.
Cheers,
Hans
For new threads USE THIS: textmate@lists.macromates.com (threading gets destroyed and the universe will collapse if you don't) http://lists.macromates.com/mailman/listinfo/textmate
On 01.02.2008, at 23:29, Peter Cowan wrote:
As for the R.app console, I didn't use it very much and find the daemon much nicer to work with.
OK
Using this version of the bundle when I use "send selection to R.app" I get the following error:
115:116: syntax error: Expected end of line, etc. but found """. (-2741)
See R.bundle help > Troubleshooting
On another note, I thought the shortcut for that command used to be shift-command-R. It's hard to remember this changed and I find the new one very difficult to type.
OK
--Hans
Hans-Joerg,
Thanks for the work on this.. After having installed the 30th jan version of the 3 R programs as instructed below... Within the R bundle (which is my default work environment), when I have a function of a loaded library, & try the "execute & insert in text" I get the following error:
library(Epi) hvgfupb<-read.csv(file="hvgfupbBackup.csv",header=TRUE) stat.table(hvgfupb$maritalcat) Error: could not find function "stat.table" Execution halted
Yet the result of the function call shows up in the "show in tooltip" version of the same command.
Also, it runs in R.app when executed in that version..
Any ideas of reasons for this behavour?
I am running Tiger (10.4.11) on an intel mac, R version 2.5.0 RC (2007-04-18 r41228)
Thanks in advance..
danstan
On Jan 30, 2008, at 1:42 PM, Hans-Joerg Bibiko wrote:
Dear R users,
in the last weeks I've looked for a better R integration into TM.
Finally I came up with these three bundles:
- R
In principal I rewrote all commands to be more flexible and to avoid to call R to get some
- R Console (Rdaemon)
I
- R Console (R.app)
In that bundle you will find all commands to remote the standard R.app. rt is to read the help file and have a look at the bundle menus.
!!VERY IMPORTANT!! Before installing these bundles please backup the 'old' R bundle (maybe, what I did is a flop ;) and delete the bundles R and - if it exists - Rdaemon; quit TM; start TM; and then install the new bundles.
All these bundles have beta status. Thus I would be very appreciated if someone could test them and give me any feedback regarding: -what doesn't work -what is illogical or unclear -bad arrangement or key binding -what's missing/wrong/unclear in the help file ...
plus what Mac OSX version, ppc or intel, which R version
If Allan is back I will upload the new R bundle to the repository and both Console bundles to the test repository.
Cheers,
Hans
<R.tmbundle.zip><R Console (Rdaemon).tmbundle.zip><R Console (R_app).tmbundle.zip>
For new threads USE THIS: textmate@lists.macromates.com (threading gets destroyed and the universe will collapse if you don't) http://lists.macromates.com/mailman/listinfo/textmate
Hi,
sorry for the delay.
On 04.02.2008, at 23:30, xolela@mac.com wrote:
Thanks for the work on this.. After having installed the 30th jan version of the 3 R programs as instructed below... Within the R bundle (which is my default work environment), when I have a function of a loaded library, & try the "execute & insert in text" I get the following error:
Is this true for ALL functions loaded e.g.: library("lattice") Depth <- equal.count(quakes$depth, number=8, overlap=.1) xyplot(lat ~ long | Depth, data = quakes)
or only for
library(Epi) hvgfupb<-read.csv(file="hvgfupbBackup.csv",header=TRUE) stat.table(hvgfupb$maritalcat)
?
I only have R 2.6.1.
If I run
library(Epi) data(warpbreaks) stat.table(tension,list(count(),mean(breaks)),data=warpbreaks)
everything works fine.
The error message says that the library(Epi) was not found.
What does happen if you run
library(Epi) hvgfupb<-read.csv(file="hvgfupbBackup.csv",header=TRUE) stat.table(hvgfupb$maritalcat)
in the Terminal running R?
Cheers,
Hans
-
Charilaos Skiadas -
Hans-Joerg Bibiko -
Hans-Jörg Bibiko
-
Michael Sheets -
Peter Cowan -
xolela＠mac.com