Problems with vi-goto-mark and vi-set-mark

Problems with vi-goto-mark and vi-set-mark

[Felix Rosencrantz]

I've recently been trying to use viins/vicmd mode.  I've hit a small
problem with the vi-set-mark and vi-goto-mark, they seem to be broken
in the latest version of zsh.  Though it looks like it was working in
4.2.5.

To reproduce:
zsh -f
% bindkey -v
% echo <ESC>maiabcdef
   # ESC changes to vicmd mode
   # ma    creates mark 'a' at the position
   # i   go back to viins mode
   # abcdef    actually doing self-inserts
% echo <ESC>`a
    # <ESC> changes to vicmd mode
    # `a   is suppose to go to mark 'a' from the previous line, but in
the latest version it doesn't seem to work.  It look like it has been
broken for more than a year in 4.3.0

-FR.

Re: Problems with vi-goto-mark and vi-set-mark

[William Scott]

Felix Rosencrantz wrote:
> I've recently been trying to use viins/vicmd mode.  I've hit a small
> problem with the vi-set-mark and vi-goto-mark, they seem to be broken
> in the latest version of zsh.  Though it looks like it was working in
> 4.2.5.
>
> To reproduce:
> zsh -f
> % bindkey -v
> % echo <ESC>maiabcdef
>    # ESC changes to vicmd mode
>    # ma    creates mark 'a' at the position
>    # i   go back to viins mode
>    # abcdef    actually doing self-inserts
> % echo <ESC>`a
>     # <ESC> changes to vicmd mode
>     # `a   is suppose to go to mark 'a' from the previous line, but in
> the latest version it doesn't seem to work.  It look like it has been
> broken for more than a year in 4.3.0
>
> -FR.
>


I just verified it works in 4.2.3 (which comes with Apple OS X) and not
with my 4.3.2

Re: Problems with vi-goto-mark and vi-set-mark

[fREW]

On 4/15/07, William Scott <wgscott@chemistry.ucsc.edu> wrote:
>
>
> Felix Rosencrantz wrote:
> > I've recently been trying to use viins/vicmd mode.  I've hit a small
> > problem with the vi-set-mark and vi-goto-mark, they seem to be broken
> > in the latest version of zsh.  Though it looks like it was working in
> > 4.2.5.
> >
> > To reproduce:
> > zsh -f
> > % bindkey -v
> > % echo <ESC>maiabcdef
> >    # ESC changes to vicmd mode
> >    # ma    creates mark 'a' at the position
> >    # i   go back to viins mode
> >    # abcdef    actually doing self-inserts
> > % echo <ESC>`a
> >     # <ESC> changes to vicmd mode
> >     # `a   is suppose to go to mark 'a' from the previous line, but in
> > the latest version it doesn't seem to work.  It look like it has been
> > broken for more than a year in 4.3.0
> >
> > -FR.
> >
>
>
> I just verified it works in 4.2.3 (which comes with Apple OS X) and not
> with my 4.3.2
>
>

Also broken in Gentoo, 4.3.2.

Re: Problems with vi-goto-mark and vi-set-mark

[Peter Stephenson]

"Felix Rosencrantz" wrote:
> I've recently been trying to use viins/vicmd mode.  I've hit a small
> problem with the vi-set-mark and vi-goto-mark, they seem to be broken
> in the latest version of zsh.  Though it looks like it was working in
> 4.2.5.

Yes, I don't use vi mode so don't notice this sort of thing.  We could
do with someone writing interactive tests for zle commands.  We could
also do with someone to look after vi mode.  We could do with people
doing a lot of things.

The change is that getchar() used not to set the last character, which
was done at a higher level, but the replacements which support multibyte
do (even if multibyte mode isn't compiled in).  This means the character
read is always treated as if it was the same as the last character.

I think the following fixes it, but I'm not sure what the feature is in
vi-goto-mark that has the special behaviour if the last character is
repeated.  If you know and can test it, please do.

Index: Src/Zle/zle.h
===================================================================
RCS file: /cvsroot/zsh/zsh/Src/Zle/zle.h,v
retrieving revision 1.34
diff -u -r1.34 zle.h
--- Src/Zle/zle.h	3 Dec 2006 21:07:18 -0000	1.34
+++ Src/Zle/zle.h	15 Apr 2007 19:22:16 -0000
@@ -72,6 +72,7 @@
 #define ZC_toupper towupper
 
 #define LASTFULLCHAR	lastchar_wide
+#define LASTFULLCHAR_T  ZLE_INT_T
 
 #else  /* Not MULTIBYTE_SUPPORT: old single-byte code */
 
@@ -130,6 +131,7 @@
 #define ZC_toupper tuupper
 
 #define LASTFULLCHAR	lastchar
+#define LASTFULLCHAR_T	int
 
 #endif
 
Index: Src/Zle/zle_move.c
===================================================================
RCS file: /cvsroot/zsh/zsh/Src/Zle/zle_move.c,v
retrieving revision 1.8
diff -u -r1.8 zle_move.c
--- Src/Zle/zle_move.c	1 Nov 2005 02:50:29 -0000	1.8
+++ Src/Zle/zle_move.c	15 Apr 2007 19:22:16 -0000
@@ -484,9 +484,10 @@
 vigotomark(UNUSED(char **args))
 {
     ZLE_INT_T ch;
+    LASTFULLCHAR_T lfc = LASTFULLCHAR;
 
     ch = getfullchar(0);
-    if (ch == LASTFULLCHAR)
+    if (ch == lfc)
 	ch = 26;
     else {
 	if (ch < ZWC('a') || ch > ZWC('z'))

-- 
Peter Stephenson <p.w.stephenson@ntlworld.com>
Web page now at http://homepage.ntlworld.com/p.w.stephenson/

Re: Problems with vi-goto-mark and vi-set-mark

[fREW]

On 4/15/07, Peter Stephenson <p.w.stephenson@ntlworld.com> wrote:
> "Felix Rosencrantz" wrote:
> > I've recently been trying to use viins/vicmd mode.  I've hit a small
> > problem with the vi-set-mark and vi-goto-mark, they seem to be broken
> > in the latest version of zsh.  Though it looks like it was working in
> > 4.2.5.
>
> Yes, I don't use vi mode so don't notice this sort of thing.  We could
> do with someone writing interactive tests for zle commands.  We could
> also do with someone to look after vi mode.  We could do with people
> doing a lot of things.
>
> The change is that getchar() used not to set the last character, which
> was done at a higher level, but the replacements which support multibyte
> do (even if multibyte mode isn't compiled in).  This means the character
> read is always treated as if it was the same as the last character.
>
> I think the following fixes it, but I'm not sure what the feature is in
> vi-goto-mark that has the special behaviour if the last character is
> repeated.  If you know and can test it, please do.
>
> Index: Src/Zle/zle.h
> ===================================================================
> RCS file: /cvsroot/zsh/zsh/Src/Zle/zle.h,v
> retrieving revision 1.34
> diff -u -r1.34 zle.h
> --- Src/Zle/zle.h       3 Dec 2006 21:07:18 -0000       1.34
> +++ Src/Zle/zle.h       15 Apr 2007 19:22:16 -0000
> @@ -72,6 +72,7 @@
>  #define ZC_toupper towupper
>
>  #define LASTFULLCHAR   lastchar_wide
> +#define LASTFULLCHAR_T  ZLE_INT_T
>
>  #else  /* Not MULTIBYTE_SUPPORT: old single-byte code */
>
> @@ -130,6 +131,7 @@
>  #define ZC_toupper tuupper
>
>  #define LASTFULLCHAR   lastchar
> +#define LASTFULLCHAR_T int
>
>  #endif
>
> Index: Src/Zle/zle_move.c
> ===================================================================
> RCS file: /cvsroot/zsh/zsh/Src/Zle/zle_move.c,v
> retrieving revision 1.8
> diff -u -r1.8 zle_move.c
> --- Src/Zle/zle_move.c  1 Nov 2005 02:50:29 -0000       1.8
> +++ Src/Zle/zle_move.c  15 Apr 2007 19:22:16 -0000
> @@ -484,9 +484,10 @@
>  vigotomark(UNUSED(char **args))
>  {
>      ZLE_INT_T ch;
> +    LASTFULLCHAR_T lfc = LASTFULLCHAR;
>
>      ch = getfullchar(0);
> -    if (ch == LASTFULLCHAR)
> +    if (ch == lfc)
>         ch = 26;
>      else {
>         if (ch < ZWC('a') || ch > ZWC('z'))
>
> --
> Peter Stephenson <p.w.stephenson@ntlworld.com>
> Web page now at http://homepage.ntlworld.com/p.w.stephenson/
>

Do you have any examples of how one could make zle unit tests?  I
would gladly do it depending on how much time I end up having this
summer.

-fREW

Re: Problems with vi-goto-mark and vi-set-mark

[Peter Stephenson]

fREW wrote:
> Do you have any examples of how one could make zle unit tests?  I
> would gladly do it depending on how much time I end up having this
> summer.

Thanks for the offer.

It should "simply" be a case of taking the code that tests completion
and making a version for general editing tests.  See Test/comptest for
the functions used and the Test/Y* tests for how the functions in
comptest are used.

Although the results should be simpler than comptest, it might not be
very clear what's going on in comptest and hence what needs taking out.
The zpty calls are the key to it; they interact with a subshell in a
pseudoterminal, which where the tests run.  A lot of the rest is
configuration to make the results visible (which will still be needed).

Any basically working system for general zle tests, even without the
tests themselves, would be extremely good to have.

-- 
Peter Stephenson <p.w.stephenson@ntlworld.com>
Web page now at http://homepage.ntlworld.com/p.w.stephenson/

Re Google Summer of Code....

[Felix Rosencrantz]

On 4/15/07, Peter Stephenson <p.w.stephenson@ntlworld.com> wrote:
> Yes, I don't use vi mode so don't notice this sort of thing.  We could
> do with someone writing interactive tests for zle commands.  We could
> also do with someone to look after vi mode.  We could do with people
> doing a lot of things.

It's looks like it is too late for 2007, but maybe we could get a 2008
SOC intern to help with zsh. I mentioned this about a year ago.  I
know in past years, they've had an intern work on Bash.

Also,  Zsh has had a strong history of student participation.  It was originally
written by Paul Falstad while as student at Princeton.  I believe
Sven worked on the bulk of the new completion system while being a student.
I'm sure there have been other serious student contributors who have
helped, whose
contribution was significant, though I can't name them.

http://code.google.com/soc/

One of the things we would need to do is create a list of project
ideas (if one doesn't already exist). Which might be useful in general
for folks who would be interesting in helping out.   It seems like
there might be interest in helping from folks

-FR.

Re: Problems with vi-goto-mark and vi-set-mark

[Felix Rosencrantz]

Thanks for the patch, Peter.  It seems to be working better now.

Though there does seem to be some oddness when trying multiple
vi-goto-mark-line  and then doing a vi-goto-mark for the same mark.
Though not sure, will need to come up with a reproducible case.

-FR.
On 4/15/07, Peter Stephenson <p.w.stephenson@ntlworld.com> wrote:
> "Felix Rosencrantz" wrote:
> > I've recently been trying to use viins/vicmd mode.  I've hit a small
> > problem with the vi-set-mark and vi-goto-mark, they seem to be broken
> > in the latest version of zsh.  Though it looks like it was working in
> > 4.2.5.
>
> Yes, I don't use vi mode so don't notice this sort of thing.  We could
> do with someone writing interactive tests for zle commands.  We could
> also do with someone to look after vi mode.  We could do with people
> doing a lot of things.
>
> The change is that getchar() used not to set the last character, which
> was done at a higher level, but the replacements which support multibyte
> do (even if multibyte mode isn't compiled in).  This means the character
> read is always treated as if it was the same as the last character.
>
> I think the following fixes it, but I'm not sure what the feature is in
> vi-goto-mark that has the special behaviour if the last character is
> repeated.  If you know and can test it, please do.
>
> Index: Src/Zle/zle.h
> ===================================================================
> RCS file: /cvsroot/zsh/zsh/Src/Zle/zle.h,v
> retrieving revision 1.34
> diff -u -r1.34 zle.h
> --- Src/Zle/zle.h       3 Dec 2006 21:07:18 -0000       1.34
> +++ Src/Zle/zle.h       15 Apr 2007 19:22:16 -0000
> @@ -72,6 +72,7 @@
>  #define ZC_toupper towupper
>
>  #define LASTFULLCHAR   lastchar_wide
> +#define LASTFULLCHAR_T  ZLE_INT_T
>
>  #else  /* Not MULTIBYTE_SUPPORT: old single-byte code */
>
> @@ -130,6 +131,7 @@
>  #define ZC_toupper tuupper
>
>  #define LASTFULLCHAR   lastchar
> +#define LASTFULLCHAR_T int
>
>  #endif
>
> Index: Src/Zle/zle_move.c
> ===================================================================
> RCS file: /cvsroot/zsh/zsh/Src/Zle/zle_move.c,v
> retrieving revision 1.8
> diff -u -r1.8 zle_move.c
> --- Src/Zle/zle_move.c  1 Nov 2005 02:50:29 -0000       1.8
> +++ Src/Zle/zle_move.c  15 Apr 2007 19:22:16 -0000
> @@ -484,9 +484,10 @@
>  vigotomark(UNUSED(char **args))
>  {
>      ZLE_INT_T ch;
> +    LASTFULLCHAR_T lfc = LASTFULLCHAR;
>
>      ch = getfullchar(0);
> -    if (ch == LASTFULLCHAR)
> +    if (ch == lfc)
>         ch = 26;
>      else {
>         if (ch < ZWC('a') || ch > ZWC('z'))
>
> --
> Peter Stephenson <p.w.stephenson@ntlworld.com>
> Web page now at http://homepage.ntlworld.com/p.w.stephenson/
>

Re: Re Google Summer of Code....

[Peter Stephenson]

"Felix Rosencrantz" wrote:
> It's looks like it is too late for 2007, but maybe we could get a 2008
> SOC intern to help with zsh. I mentioned this about a year ago.  I
> know in past years, they've had an intern work on Bash.

It would be useful, although the zsh code is quite difficult to come to
cold, particularly the areas that really need the work.  The project
would probably need quite a lot of supervision.

-- 
Peter Stephenson <p.w.stephenson@ntlworld.com>
Web page now at http://homepage.ntlworld.com/p.w.stephenson/