fix readline extra space issue in sage (python vs. readline problem)

[williamstein]

>  > On Mon, Mar 10, 2008 at 1:27 PM, Nikos Apostolakis <[email protected]> wrote:
>  >  >
>  >  >  Hello group,
>  >  >
>  >  >  the following is not terribly important but it bothers me.  When
>  >  >  using the cli to SAGE the completion adds extra space after the
>  >  >  name of functions or methods. For example if I hit den[tab] in the
>  >  >  sage prompt I get denominator *, where the star stands for the
>  >  >  cursor, that is there is an extra space after "denominator".   I am
>  >  >  then forced to either have a space between a function and the
>  >  >  parenthesis enclosing its arguments or to move the cursor back one
>  >  >  space.  The former annoys me aesthetically and the latter wastes my
>  >  >  time (albeit a small fraction of it).
>  >  >
>  >  >  Is there a way to turn this off? some configuration option perhaps?
>  >  >
>  >
>  >  (1) This is an Ipython problem.  Probably Sage itself has little to do
>  >  with it.  Hence I've cc'd this message to Fernando Perez (author
>  >  of IPython).
>  >
>  >  (2) I just checked on a bunch of machines.  The behavior you mention
>  >  above does *not* occur on any of my OS X machines.  It also does
>  >  not occur for me on sage.math (a Debian 64-bit machine).  It does
>  >  occur on several 32-bit and 64-bit Linux installs that I have handy.
>  >  Sage builds its own readline and uses exact the same config files
>  >  on all machines, so I'm fairly puzzled by all this.
>
>  I'm afraid it's not really an ipython problem, but how readline ends
>  up built on any given box.  I've fought this particular one before:
>
>  http://projects.scipy.org/pipermail/ipython-user/2005-April/002617.html
>
>  As you can see, the python readline doesn't expose the necessary
>  functionality at all:
>
>  http://mail.python.org/pipermail/python-list/2002-December/176942.html
>
>  So the actual behavior will be hardcoded by however readline ends up
>  built.  You may be able to tweak *your* readline build so that
>  HAVE_RL_COMPLETION_APPEND_CHARACTER is defined at build time and hence
>  you get the null completion character.

Excellent.  Since Sage builds its own readline, we can do what it takes
to make sure this works.

Thanks!

Component: user interface

Issue created by migration from https://trac.sagemath.org/ticket/2469

[jasongrout]

comment:1

Another ticket for this issue is at #2306

[sagetrac-mabshoff]

comment:2

I just checked and on all my builds I get the following:

sage-3.1.3.alpha2/local/include$ grep -r HAVE_RL_COMPLETION_APPEND_CHARACTER *
python2.5/pyconfig.h:#define HAVE_RL_COMPLETION_APPEND_CHARACTER 1

In #3947 we fixed the python.spkg LDFLAGS to pick up Sage's python, so I consider this issue resolved. I checked about 8 different Linux builds and cannot reproduce this problem. So unless someone comes up with a broken system I am closing this as fixed.

Cheers,

Michael