[ESS-bugs] ess 5.2.2; TAB completion takes *ages* ..

[Martin Maechler]

Yes indeed, this has not been a problem on Linux with R.

You indicated using Emacs 20.7 which is from the last century.
I don't really believe the problem to be an Emacs 20 one, but
since ``everyone else'' uses Emacs 21.3 (I use 21.2
occasionally) it would definitely make debugging easier.

Thanks a lot for offering to debug this problem.
AFAICS, I have no ess-r-namedb.el file either (on Linux) and
still things work very fast.

What happens when you (go to the *R* buffer and) type
M-x ess-resynch ?
It takes about half a second for me (fast Linux box though)
your "current state" (below) looks exactly like mine...

Regards,
Martin

>>>>> "David" == David Heffernan <david at heffs.org.uk>
>>>>>     on Wed, 18 Aug 2004 20:50:12 +0000 writes:

    David> Hi, I'm having problems with ESS tab completion.  The
    David> problem occurs on both Windows 2000 and XP and
    David> possible other Windows versions.  I'm using R version
    David> 1.8.1 but I've had the problem with a various earlier
    David> versions too so I don't think the R version is a
    David> significant factor.

    David> The first time you attempt a tab completion in ESS it
    David> forms the list of available completions.  This
    David> operation takes a *very* long time.  After that
    David> completion works a treat for the duration of the ESS
    David> session.  When you start another session it's back to
    David> square one.

    David> I'm aware of the ess-object-name-db game.  I followed
    David> the instructions to create the ess-r-namedb.el file.
    David> Then when I start ESS again the ess-object-name-db
    David> variable is indeed defined (it's a local to the *R*
    David> buffer).  But still the first tab completion takes an
    David> age.

    David> I know that there a many other users suffering with
    David> the same problem (this is clear as a result of my
    David> websearches to attempt to find a solution).  Do you
    David> know of a solution to the problem?

    David> The fact that the problem has been known for a while
    David> and there has been no solution makes me wonder
    David> whether the ESS developers can reproduce it.  I know
    David> Linux users who don't experience this problem so that
    David> seems plausible.  If this is the case then perhaps I
    David> can help track it down since I can easily reproduce
    David> the problem.  I work as a software developer and so
    David> am experienced in tracking and debugging software
    David> problems.  However, I've never debugged anything like
    David> Emacs or ESS and realise that I would need some help
    David> setting up a dev environment for debugging.


    David> Cheers,


    David> David Heffernan.


    David> Emacs  : GNU Emacs 20.7.1 (i386-*-nt5.1.2600)
    David> of Tue Jun 13 2000 on buffy
    David> Package: ess-mode 5.2.2

    David> current state:
    David> ==============
    David> (setq
    David> ess-language "Initial"
    David> ess-dialect nil
    David> ess-ask-for-ess-directory t
    David> ess-ask-about-transfile nil
    David> ess-directory nil
    David> ess-keep-dump-files "always"
    David> ess-source-directory "/tmp/"
    David> )
    David> [ess-site.el]: ess-customize-alist=nil
    David> [ess-site.el _2_]: ess-customize-alist=nil
    David> (R): ess-rterm-versions-create making M-x defuns for