[ESS-bugs] ess-function-pattern not working for some valid functions

Tue Mar 15 15:54:19 CET 2011

Dear Martin,

Toby and I have already discovered our coincident interest in this and have been in communication, with the upshot that we will put the revised regexps on the inlinedocs web site. I appreciate your reasons for caution.

It turns out that neither of us had our original submissions quite right (among other things we also need a change to ess-set-function-pattern), but I have been happily working with the current version for some days now. However I am not a heavy user of S4 classes so have only done limited tests on the "set" functions. Is there a "torture test" for these function patterns somewhere? 

>From a long term maintenance point of view, I would still like to propose a change to the ESS sources to avoid forking - for example I expect you will be modifying or have modified to support setRefClass - but I would be happy to accept some kind of "quarantine period" while we (and I hope some others) try out the patterns.

My current ess-mode-hook includes the following, but I think that I could remove everything after function\\b, so will try that out when I have time:

(add-hook 'ess-mode-hook
 '(lambda()
    ;; modifications to function pattern allow R.oo setConstructorS3, setMethodS3, (inlinedocs) structure call and documentation between keyword function and opening (
    (setq ess-function-pattern "\\(\\(\\(\\s\"\\[?\\[?\\(\\sw\\|\\s_\\)*\\(<-\\)?\\(\\sw\\|\\s_\\)*\\s\"\\)\\|\\(\\(^\\|[ ]\\)\\(\\sw\\|\\s_\\)+\\)\\)\\s-*\\(<-\\|=\\)\\|^set\\(As\\|Method\\|MethodS3\\|ConstructorS3\\|Generic\\|GroupMethod\\|ReplaceMethod\\)(\\s\"\\[?\\[?\\(\\sw\\|\\s_\\)*\\s\",\\(\\s-\\|\n\\)*.*\\)\\(\\(\\s-\\|\n\\)*\\s<.*\\s>\\)*\\(\\s-\\|\n\\)*\\(structure\\s-*(\\s-*\\(?:\\(?:#.*\\)?\n\\s-*\\)*\\)?function\\b\\s-*\\(?:\\(?:#.*\\)?\n\\s-*\\)*(?")
    ;; modifications to function start allow R.oo setConstructorS3, setMethodS3 and (inlinedocs) structure call
    (setq ess-set-function-start "\\(^set[MGARC][Ma-z]+\\(?:S3\\)?\\)\\|\\(.*<-\\s-*structure\\)\\s-*(")
    ))

Regards,

Keith

P.S. I forgot to say last time how much I appreciate ESS and all the work you do - I am an avid lurker on the ess-help list!

Keith Ponting
Aurix Ltd, Malvern WR14 3SZ  UK

> -----Original Message-----
> From: Martin Maechler [mailto:maechler at stat.math.ethz.ch]
> Sent: 15 March 2011 14:01
> To: Toby Dylan Hocking
> Cc: Keith Ponting; ess-bugs at stat.math.ethz.ch
> Subject: Re: [ESS-bugs] ess-function-pattern not working for some valid
> functions
> 
> Dear Toby,
> your attachment did not come through the filters
>      .. -> .. -> ESS-bugs -> ESS-core ->
> 
> but interestingly, somewhat simultaneously to your post to ESS-bugs,
> there was one to ESS-help from Keith Ponting.
> see the list archive entry
>     https://stat.ethz.ch/pipermail/ess-help/2011-March/006764.html
>     (and then the attachment in there -- he used HTML in e-mail --)
> 
> where he writes
> 
>  > However the comments between the function keyword and the opening  >
> parenthesis do not match ess-function-pattern. I realise that it is  >
> already an extremely complicated regexp, but after a bit of tweaking I
> > found that the following change (which also allows the function
> argument  > list opening parenthesis to be on a new line without any
> intervening  > comments) appears to work: replace the trailing
> "function\\s-*(" with  > "function\\s-*\\(?:\\(?:#.*\\)?\n\\s-*\\)*(".
> This modification can be  > done locally in any number of ways (emacs
> file-local variables,  > directory-local variables, ess-mode-hook etc.),
> but I wonder whether I  > could propose the above as a modification to
> ESS?
> 
> and I haven't much tried his proposition yet, either.
> 
> I do know that I had spent too many numerous hours in the past, fiddling
> with those regexps, and the last few improvement attempts all had failed
> in sofar as they  fixed one case  but  broke another one.
> 
> Have you guys both worked for several days with the *changed* regexps?
> (I don't know yet Toby's changes).
> 
> Toby, could you show us your patch?
> 
> Thank you for your contributions; I'd definitely like to look into.
> 
> Best regards,
> Martin
> 
> Martin <Maechler at stat.math.ethz.ch>  http://stat.ethz.ch/people/maechler
> Seminar für Statistik, ETH Zürich  HG G 16      Rämistrasse 101
> CH-8092 Zurich, SWITZERLAND
> phone: +41-44-632-3408       fax: ...-1228      <><
> 
> 
> >>>>> Toby Dylan Hocking <Toby.Hocking at inria.fr>
> >>>>>     on Thu, 10 Mar 2011 15:00:33 +0100 writes:
> 
>     > Dear ESS developers,
>     > First of all, thank you very much for ESS, a great piece of free
>     > software which I use every day!
> 
>     > I like using the C-c C-f key sequence to load individual
>     > functions into an inferior R process, but I noticed that it
>     > doesn't work for some syntactically-valid functions, such as
> 
>     > comments.and.ex <- structure(function
>     > ### some docs
>     > ### more
>     > (arg,
>     > arg2
>     > ){
>     > foo
>     > ### value
>     > },ex=function(){
>     > some.code
>     > })
> 
>     > To solve this problem, I propose a patch to ess-mode.el, which
>     > simply updates the regular expression used to match functions.
> 
>     > I have attached the patch and some R code that illustrates
>     > several example functions which are valid R code, but are not
>     > currently matched by ess-function-pattern.
> 
>     > Does this patch pose any problem for you guys? Can you
>     > incorporate it into the main ESS source code?
> 
>     > Thanks for the great package, and for your consideration,
>     > Toby Dylan Hocking
>     > http://cbio.ensmp.fr/~thocking/
> 
> 
>     > ------------------------------------------------------------------
> ----
>     > testfun <- function(foo){ bar } testfun <- function (foo){ bar }
>     > oneline <- function(foo)bar comments <- function ### some docs
>     > ### more (arg, arg2 ){ foo
>     > ###
>     > }
> 
>     > comments.and.ex <- structure(function
>     > ### some docs
>     > ### more
>     > (arg,
>     > arg2
>     > ){
>     > foo
>     > ###
>     > },ex=function(){
>     > some.code
>     > })
> 
>     > ------------------------------------------------------------------
> ----
>     > _______________________________________________ ESS-bugs
>     > ESS-bugs at r-project.org
>     > https://stat.ethz.ch/mailman/listinfo/ess-bugs
> 
>     > ------------------------------------------------------------------
> ----
>     > _______________________________________________ ESS-core list:
>     > https://stat.ethz.ch/mailman/listinfo/ess-core