fixsuffix() again

fixsuffix() again

[Bart Schaefer]

Sven, one of your patch hunks in zsh-workers/4510 removes a call to
fixsuffix() from the top of docompletion().  Is this intentional or not?

(I ask because that same patch also backs out the fixsuffix() from 4412,
which you then had to put back, slightly relocated, in 4528.)

-- 
Bart Schaefer                                 Brass Lantern Enterprises
http://www.well.com/user/barts              http://www.brasslantern.com

Re: fixsuffix() again

[Bart Schaefer]

On Nov 6, 10:11am, Sven Wischnowsky wrote:
} Subject: Re: fixsuffix() again
}
} "Bart Schaefer" wrote:
} 
} > So you mean with that one call to fixsuffix() in docompletion() we can
} > get rid of all the other calls to fixsuffix() scattered about in other
} > places?  If so, then by golly I vote for putting that one back and
} > nuking the others.
} 
} The one in deletecharorlist() will still be need since it is in the
} branch where completion is noct called.   [...]   With
} fixsuffix() in docompletion() the slash is not removed, but with the
} old style it is.

It's not worth it, then.  I thought you meant you could get rid of at
least five of the other six calls that way.

-- 
Bart Schaefer                                 Brass Lantern Enterprises
http://www.well.com/user/barts              http://www.brasslantern.com

Re: fixsuffix() again

[Sven Wischnowsky]

Peter Stephenson wrote:

> 
> Sven Wischnowsky wrote:
> > > So you mean with that one call to fixsuffix() in docompletion() we can
> > > get rid of all the other calls to fixsuffix() scattered about in other
> > > places?  If so, then by golly I vote for putting that one back and
> > > nuking the others.
> > 
> > The one in deletecharorlist() will still be need since it is in the
> > branch where completion is noct called. So it would make two instead
> > of three calls in zle_tricky.c. The patch below does this, but note
> > that there is one ugliness:
> 
> I've found a more serious problem with this.
> 
> % export ZMLIB=$PWD               # some people probably have this set anyway
> % setopt noautomenu menucomplete noautolist
> % echo $Z<TAB>            -> $ZMLIST/
>          <TAB>            -> $ZSH_NAME
>          <TAB>            -> $ZSH_VERSION
>          <RET>            -> $ZSH_VERSIO
> 
> It seems to lose memory of the fact that the slash has already gone.

Ok, then there is another reason not to move the fixsuffix()
there (the problem Peter reported can be fixed by the fixsuffix() in
do_single()).


Bye
 Sven


--
Sven Wischnowsky                         wischnow@informatik.hu-berlin.de

Re: fixsuffix() again

[Peter Stephenson]

Sven Wischnowsky wrote:
> > So you mean with that one call to fixsuffix() in docompletion() we can
> > get rid of all the other calls to fixsuffix() scattered about in other
> > places?  If so, then by golly I vote for putting that one back and
> > nuking the others.
> 
> The one in deletecharorlist() will still be need since it is in the
> branch where completion is noct called. So it would make two instead
> of three calls in zle_tricky.c. The patch below does this, but note
> that there is one ugliness:

I've found a more serious problem with this.

% export ZMLIB=$PWD               # some people probably have this set anyway
% setopt noautomenu menucomplete noautolist
% echo $Z<TAB>            -> $ZMLIST/
         <TAB>            -> $ZSH_NAME
         <TAB>            -> $ZSH_VERSION
         <RET>            -> $ZSH_VERSIO

It seems to lose memory of the fact that the slash has already gone.

-- 
Peter Stephenson <pws@ibmth.df.unipi.it>       Tel: +39 050 844536
WWW:  http://www.ifh.de/~pws/
Dipartimento di Fisica, Via Buonarotti 2, 56100 Pisa, Italy

Re: fixsuffix() again

[Sven Wischnowsky]

"Bart Schaefer" wrote:

> On Nov 6,  9:16am, Sven Wischnowsky wrote:
> } Subject: Re:  fixsuffix() again
> }
> } Bart Schaefer wrote:
> } > Sven, one of your patch hunks in zsh-workers/4510 removes a call to
> } > fixsuffix() from the top of docompletion().  Is this intentional or not?
> } 
> } I have no idea where this fixsuffix() in docompletion() comes from
> 
> Looks like Zefram added it when he did the 3.1.5 release.  It's not in
> any version I have in CVS until after I import the 3.1.5 code and merge.
> 
> } As I wrote in another message this may be a issue of personal
> } taste. If we put a fixsuffix() there any suffix on the line that is
> } autoremovable will become fixed when completion is tried in any way
> } (listing matches, completion with no matches, ...).
> 
> So you mean with that one call to fixsuffix() in docompletion() we can
> get rid of all the other calls to fixsuffix() scattered about in other
> places?  If so, then by golly I vote for putting that one back and
> nuking the others.

The one in deletecharorlist() will still be need since it is in the
branch where completion is noct called. So it would make two instead
of three calls in zle_tricky.c. The patch below does this, but note
that there is one ugliness:

In the main zsh directory do:

  ls Sr<TAB>

giving you

  ls Src/

now type C-d, you get the matches in Src, then type a space. With
fixsuffix() in docompletion() the slash is not removed, but with the
old style it is.

Bye
 Sven

*** os/Zle/zle_tricky.c	Fri Nov  6 10:09:42 1998
--- Src/Zle/zle_tricky.c	Fri Nov  6 10:10:19 1998
***************
*** 2927,2932 ****
--- 2927,2934 ----
  static void
  docompletion(char *s, int lst, int incmd)
  {
+     fixsuffix();
+ 
      HEAPALLOC {
  	pushheap();
  
***************
*** 4807,4814 ****
  	if (!ainfo)
  	    return;
  
- 	fixsuffix();
- 
  	/* Delete the old stuff from the command line. */
  	cs = wb;
  	foredel(we - wb);
--- 4809,4814 ----
***************
*** 4938,4945 ****
      if (!prpre) prpre = "";
      if (!ppre) ppre = "";
      if (!psuf) psuf = "";
- 
-     fixsuffix();
  
      if (!menucur) {
  	/* We are currently not in a menu-completion, *
--- 4938,4943 ----


--
Sven Wischnowsky                         wischnow@informatik.hu-berlin.de

Re: fixsuffix() again

[Bart Schaefer]

On Nov 6,  9:16am, Sven Wischnowsky wrote:
} Subject: Re:  fixsuffix() again
}
} Bart Schaefer wrote:
} > Sven, one of your patch hunks in zsh-workers/4510 removes a call to
} > fixsuffix() from the top of docompletion().  Is this intentional or not?
} 
} I have no idea where this fixsuffix() in docompletion() comes from

Looks like Zefram added it when he did the 3.1.5 release.  It's not in
any version I have in CVS until after I import the 3.1.5 code and merge.

} As I wrote in another message this may be a issue of personal
} taste. If we put a fixsuffix() there any suffix on the line that is
} autoremovable will become fixed when completion is tried in any way
} (listing matches, completion with no matches, ...).

So you mean with that one call to fixsuffix() in docompletion() we can
get rid of all the other calls to fixsuffix() scattered about in other
places?  If so, then by golly I vote for putting that one back and
nuking the others.

-- 
Bart Schaefer                                 Brass Lantern Enterprises
http://www.well.com/user/barts              http://www.brasslantern.com

Re: fixsuffix() again

[Sven Wischnowsky]

Bart Schaefer wrote:

> 
> Sven, one of your patch hunks in zsh-workers/4510 removes a call to
> fixsuffix() from the top of docompletion().  Is this intentional or not?
> 
> (I ask because that same patch also backs out the fixsuffix() from 4412,
> which you then had to put back, slightly relocated, in 4528.)

I have no idea where this fixsuffix() in docompletion() comes from
(looking at the various versions of 3.1.4 I have and the last few
hundred mail messages).
As I wrote in another message this may be a issue of personal
taste. If we put a fixsuffix() there any suffix on the line that is
autoremovable will become fixed when completion is tried in any way
(listing matches, completion with no matches, ...).

Bye
 Sven


--
Sven Wischnowsky                         wischnow@informatik.hu-berlin.de