[ESS-bugs] ess-mode 5.3.6; opening file with tramp mode -> error

Stephen Eglen S.J.Eglen at damtp.cam.ac.uk
Wed Oct 10 22:34:57 CEST 2007 

hi David,

thanks for the useful report -- and sorry for the delay in writing.
It still looks to me like a tramp issue.  Just this afternoon I tried
tramp again (for the first time in a couple of years).  it works but
is quite slow I find.  Although this doesn't fix the issue, have you
tried FUSE to transparently mount a remote filesystem?  (I guess it
depends how often you visit these other files through ssh).

I've got some teaching to finish this week, but will send in a report
to tramp and emacs-devel -- as I read this, it is not an ESS bug.

Stephen


David Reitter writes:
 > Stephen,
 > 
 > this year-old issue has popped up again.
 > 
 > When I start Aquamacs from the command line and then autoload R-mode  
 > when accessing an .R file via tramp, I get the "opening directory"  
 > error again, and I believe it is, once more, due to a tramp-style  
 > default-directory and an element of "." in the exec-path.
 > 
 > The debugger trace is:
 > 
 > Debugger entered--Lisp error: (file-error "Opening directory" "no  
 > such file or directory" "/ssh:xxx at ssh.xxx.ed.ac.uk:/home/xxx/PhD/stats")
 >    file-name-all-completions("Splus" ".")
 >    ess-find-exec-completions("Splus")
 >    mapcar(ess-find-exec-completions ("Splus"))
 >    ess-s-versions-create()
 >    byte-code("ƒ ,AFG(BMˆ,AHI(BMˆ,AJK(BMˆ,ALM(BMˆ‚8 	 
 > ,AN(B=ƒ, ,AFG(BMˆ,AJK(BMˆ,ALM(BMˆ‚8 ,AFG(BMˆ,AJK(BMˆ,ALM(BMˆ,AOL(BMˆ,APJ(BMˆ,AC(B\nB,AQC(B!„L ,AR(B,AR‰S(B*ƒb ,AT(B  
 > *,AU(B +‚f ,AV(B *,AW(B ,AXYZ[(B\"!‰ƒ
 >  ,AY\(B
 > \",,A](B-,A^_(B,B#ˆ)+,A`(B ˆ.ƒ  ,Aabcd(B#ˆ,Aaecd(B#ˆ,Aafcd(B#ˆ,Ad(B/,Ag(B0ƒ,A/(B ,Aahi(B\"ˆ,AR‡(B" [ess- 
 > microsoft-p system-type current-load-list ess-r-versions-created R- 
 > newest-list ess-versions-created S S+6 Sqpe Sqpe+6 s-mode S+6-mode s- 
 > transcript-mode S+6-transcript-mode gnu/linux S-transcript-mode S- 
 > mode boundp nil ("R-newest") ess-sqpe-versions-create ess-find-rterm  
 > ess-s-versions-create ess-r-versions-create ess-flatten-list mapcar # 
 > [(x) ",AA(B!
 >  J‡" [x boundp] 2] (R-newest-list ess-r-versions-created ess-s- 
 > versions-created) #[(x) ",AA(B,AB(B!,AC(B#‡" [x vector intern t] 4] easy-menu- 
 > add-item ("Start Process") "Other" ess-check-R-program-name add-hook  
 > ess-mode-hook turn-on-font-lock t ess-transcript-mode-hook inferior- 
 > ess-mode-hook "always" ess-post-run-hook (lambda nil (if ... ...))  
 > ess-s-versions-created ess-rterm-version-paths new-menu ess-mode-menu  
 > window-system inferior-ess-font-lock-input ess-keep-dump-files] 6)
 >    R-mode()
 >    set-auto-mode-0(R-mode nil)
 >    set-auto-mode()
 >    normal-mode(t)
 >    after-find-file(nil t)
 >    find-file-noselect-1(#<buffer stats2.R> "/ 
 > ssh:xxx at ssh.xxx.ed.ac.uk:/home/xxx/PhD/stats/stats2.R" nil nil "/ 
 > ssh:xxx at ssh.xxx.ed.ac.uk:/amd/nfs/xxx/disk/ptn053/xxx/PhD/stats/ 
 > stats2.R" ((67 . 58701) (-1 0)))
 >    find-file-noselect("/ssh:xxx at ssh.xxx.ed.ac.uk:/home/xxx/PhD/stats/ 
 > stats2.R" nil nil nil)
 >    find-file("/ssh:xxx at ssh.xxx.ed.ac.uk:/home/xxx/PhD/stats/stats2.R")
 > 
 > 
 > If I manipulate `ess-find-exec-completions' to show me the `default- 
 > directory', it is "/ssh:xxx at ssh.xxx.ed.ac.uk:/home/xxx/PhD/stats/".
 > The `exec-path' at the time is
 > 
 > ("/Library/Frameworks/Python.framework/Versions/Current/bin" "/opt/ 
 > local/lib/postgresql82/bin" "/opt/local/bin" "/opt/local/sbin" "/bin"  
 > "/sbin" "/usr/bin" "/usr/sbin" "/usr/local/bin" "/usr/texbin" "/Users/ 
 > dr/Applications" "/Users/dr/Applications/bin" "/Users/dr/Projects/ 
 > openccg/bin" "/usr/texbin/powerpc-apple-darwin-current" "" "/Library/ 
 > Frameworks/Python.framework/Versions/Current/bin" "/opt/local/lib/ 
 > postgresql82/bin" "/opt/local/bin" "/opt/local/sbin" "/bin" "/sbin" "/ 
 > usr/bin" "/usr/sbin" "/usr/local/bin" "/usr/texbin" "/Users/dr/ 
 > Applications" "/Users/dr/Applications/bin" "/Users/dr/Projects/ 
 > openccg/bin" "/usr/texbin/powerpc-apple-darwin-current" "" "/Library/ 
 > Frameworks/Python.framework/Versions/Current/bin" "/opt/local/lib/ 
 > postgresql82/bin" "/opt/local/bin" "/opt/local/sbin" "/bin" "/sbin" "/ 
 > usr/bin" "/usr/sbin" "/usr/local/bin" "/usr/texbin" "/Users/dr/ 
 > Applications" "/Users/dr/Applications/bin" "/Users/dr/Projects/ 
 > openccg/bin" "/usr/texbin/powerpc-apple-darwin-current" "" "." "/ 
 > Applications/Aquamacs Emacs.app/Contents/MacOS/libexec" "/ 
 > Applications/Aquamacs Emacs.app/Contents/MacOS/bin" "/usr/local/teTeX/ 
 > bin/powerpc-apple-darwin-current")
 > 
 > (I know, some of this doesn't make sense - I don't know why it has  
 > all of these double entries.)
 > 
 > 
 > - David
 > 
 > 
 > 
 > --
 > http://aquamacs.org -- Aquamacs: Emacs on Mac OS X
 > http://aquamacs.org/donate -- Could we help you? Return the favor and  
 > support the Aquamacs Project!
 > 
 > 
 > 
 > 
 > On 16 May 2006, at 17:02, Stephen Eglen wrote:
 > 
 > >
 > > Thanks David.
 > >
 > >>
 > >> To be more precise: exec-path contains a "" element. It may contain a
 > >> nil, as per doc (and if the "" was nil, you'd still get an error
 > >> there I believe):
 > >>
 > >> "*List of directories to search programs to run in subprocesses.
 > >> Each element is a string (directory name) or nil (try default
 > >> directory)."
 > >>
 > >> Now, you're going through exec-path and looking up files there.
 > >> You're calling `file-name-all-completions', in this case with a nil
 > >> argument, which leads to file-name-all-completions using the current
 > >> working dir.
 > >>
 > >> This working dir, at that point, is a tramp URL rather than a normal
 > >> file name. That's because we're autoloading ESS, and it gets loaded
 > >> just when I access a .R file over tramp. So when you call `(file- 
 > >> name-
 > >> all-completions "R-1" "")', you get the error because that function
 > >> doesn't seem to be able to deal with tramp working directories.
 > >>
 > >> Do you agree with me that this is a bug in Emacs?
 > >
 > > Hmm.  Is it a bug in Emacs or Tramp?  it sounds more like a tramp bug
 > > to me, since the doc of file-name-all-completions does nothing to
 > > suggest that a value of "" or nil for directory will convert to
 > > default-directory; that comes from exec-path.
 > >
 > > Ignoring ESS for now, if you load a remote file using tramp, what is
 > > the value of default-directory in that file's buffer?
 > >
 > >
 > >> But I would still write something like
 > >>
 > >> 	    (setq ess-tmp-dir (nth i ess-exec-path))
 > >> 	    (if (and (> (length ess-tmp-dir) 0)
 > >> 		     (file-exists-p ess-tmp-dir)) (progn
 > >>
 > >>
 > >> because you don't actually want to include whatever is the current
 > >> working directory into the list of completion items, because this
 > >> function is executed upon loading ess-site, and it shouldn't matter
 > >> what happens to be the current directory. Right?
 > >
 > > yes, I'd go along with that.  If you add that code, does it solve your
 > > autoloading problem?  If so, let me know, I'll put in, then get
 > > ess-core to test it out.
 > >
 > > Stephen
 > 
 > _______________________________________________
 > ESS-bugs ESS-bugs at stat.math.ethz.ch
 > https://stat.ethz.ch/mailman/listinfo/ess-bugs
 > 
 > _______________________________________________
 > ESS-core list: https://stat.ethz.ch/mailman/listinfo/ess-core

More information about the ESS-bugs mailing list