Re: PATCH: yet another match-spec-cursor bug

Re: PATCH: yet another match-spec-cursor bug

[Sven Wischnowsky]

I wrote:

> Felix Rosencrantz wrote:
> 
> ...
> 
> > If I try completing "bui", I rarely I get only "build" or  more often I get the
> > following sequence after hitting tab:
> >    bui --> build.out[] --> build[.]out --> build.out[] --> build.out1[]
> >          --> build.out2[]
> > 
> > Most times I get the correct behavior.  Though within several attempts I get
> > the
> > described incorrect behavior.  
> 
> That's bad news: not always repeatable. I've tried it for quite some
> time now and still can't reproduce it. I'll continue to try...
> 
> The fact that you don't always get it makes me think that it may be a
> problem with an not-initialised variable or memory block. Do you have
> configured with mem-debug?
> 
> What really irritates me is that it doesn't go into menu-completion on 
> the second TAB. Hm.

I've now tried it on DU 4.0 and Linux, both with and without mem-debug 
and still can't reproduce it. Does anyone else see this behaviour?


I'm not even completely sure where to look when trying to debug this...

Bye
 Sven


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

Re: PATCH: yet another match-spec-cursor bug

[Felix Rosencrantz]

Sven Wischnowsky wrote:
>The fact that you don't always get it makes me think that it may be a
>problem with an not-initialised variable or memory block. Do you have
>configured with mem-debug?

I typically run without --enable-zsh-mem-debug configured.  I built configured
with this enabled, and completion worked properly.  I would agree that
it seems like some sort of initialization problem.


>> host% zstyle ':completion:*:complete:*' matcher-list 'm:{a-zA-Z}={A-Za-z} 
>> r:|[.,_-]=* r:[^A-Z0-9]||[A-Z0-9]=* r:[A-Z0-9]||[^A-Z0-9]=*
r:[^0-9]||[0-9]=*'
>> host% more bui<TAB>
>Both `r:[^A-Z0-9]||[A-Z0-9]=*' and `r:[^0-9]||[0-9]=*' doesn't seem to 
>make much sense, does it?

Yep. I should have split them up: "r:[^A-Z]||[A-Z]=* r:[^0-9]||[0-9]=*"


>What really irritates me is that it doesn't go into menu-completion on 
>the second TAB. Hm.
Hm...

-FR.

__________________________________________________
Do You Yahoo!?
Send instant messages & get email alerts with Yahoo! Messenger.
http://im.yahoo.com/

Re: PATCH: yet another match-spec-cursor bug

[Sven Wischnowsky]

Felix Rosencrantz wrote:

> Sven Wischnowsky wrote:
> >I've tried quite a bit now, but can't reproduce it. And you didn't
> >give us much information, did you? ;-) For example, I guess you have a 
> >`r:[^A-Z0-9]||[A-Z0-9]=*', right?
> >
> >So, can you reproduce it with `zsh -f'?
> 
> Sorry not to provide more details.  I was thinking the problem might be
> closely related to the original patch.

It may be... although I'm not sure.

> I am able to get the same intermittent problem with "zsh -f".
> myprompt@ zsh -f
> host% mkdir j ; touch j/build.out{,1,2} ; cd j
> host% bindkey -e; autoload -U compinit; compinit -D
> host% zstyle ':completion:*:complete:*' matcher-list 'm:{a-zA-Z}={A-Za-z} 
> r:|[.,_-]=* r:[^A-Z0-9]||[A-Z0-9]=* r:[A-Z0-9]||[^A-Z0-9]=* r:[^0-9]||[0-9]=*'
> host% more bui<TAB>

Both `r:[^A-Z0-9]||[A-Z0-9]=*' and `r:[^0-9]||[0-9]=*' doesn't seem to 
make much sense, does it?

> If I try completing "bui", I rarely I get only "build" or  more often I get the
> following sequence after hitting tab:
>    bui --> build.out[] --> build[.]out --> build.out[] --> build.out1[]
>          --> build.out2[]
> 
> Most times I get the correct behavior.  Though within several attempts I get
> the
> described incorrect behavior.  

That's bad news: not always repeatable. I've tried it for quite some
time now and still can't reproduce it. I'll continue to try...

The fact that you don't always get it makes me think that it may be a
problem with an not-initialised variable or memory block. Do you have
configured with mem-debug?

What really irritates me is that it doesn't go into menu-completion on 
the second TAB. Hm.

Bye
 Sven


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

Re: PATCH: yet another match-spec-cursor bug

[Felix Rosencrantz]

Sven Wischnowsky wrote:
>I've tried quite a bit now, but can't reproduce it. And you didn't
>give us much information, did you? ;-) For example, I guess you have a 
>`r:[^A-Z0-9]||[A-Z0-9]=*', right?
>
>So, can you reproduce it with `zsh -f'?

Sorry not to provide more details.  I was thinking the problem might be
closely related to the original patch.

I am able to get the same intermittent problem with "zsh -f".
myprompt@ zsh -f
host% mkdir j ; touch j/build.out{,1,2} ; cd j
host% bindkey -e; autoload -U compinit; compinit -D
host% zstyle ':completion:*:complete:*' matcher-list 'm:{a-zA-Z}={A-Za-z} 
r:|[.,_-]=* r:[^A-Z0-9]||[A-Z0-9]=* r:[A-Z0-9]||[^A-Z0-9]=* r:[^0-9]||[0-9]=*'
host% more bui<TAB>

If I try completing "bui", I rarely I get only "build" or  more often I get the
following sequence after hitting tab:
   bui --> build.out[] --> build[.]out --> build.out[] --> build.out1[]
         --> build.out2[]

Most times I get the correct behavior.  Though within several attempts I get
the
described incorrect behavior.  

-FR


__________________________________________________
Do You Yahoo!?
Send instant messages & get email alerts with Yahoo! Messenger.
http://im.yahoo.com/

Re: PATCH: yet another match-spec-cursor bug

[Sven Wischnowsky]

Felix Rosencrantz wrote:

> Sven Wischnowsky wrote:
> >With partial word matching and matches like `c6.yo' and `c6.yo.old',
> >the cursor was left on the dot. Caused by a thinko in the code that
> >marks the cline structs to say in which of them characters were
> >missing.
> 
> I'm seeing a similar problem with the latest code.  I have a directory with
> only three files named build.out{,1,2} and sometimes I see incorrect behavior
> with completion.
> 
> If I try completing "bui", I sometimes get only "build" or more often I get the
> following sequence from hitting tab:
>    bui --> build.out[] --> build[.]out --> build.out[] --> build.out1[]
>          --> build.out2[]
> The cursor jumps back from the end to the dot, and then back to the end with
> normal behavior.

I've tried quite a bit now, but can't reproduce it. And you didn't
give us much information, did you? ;-) For example, I guess you have a 
`r:[^A-Z0-9]||[A-Z0-9]=*', right?

So, can you reproduce it with `zsh -f'?

Bye
 Sven


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

Re: PATCH: yet another match-spec-cursor bug

[Felix Rosencrantz]

Sven Wischnowsky wrote:
>With partial word matching and matches like `c6.yo' and `c6.yo.old',
>the cursor was left on the dot. Caused by a thinko in the code that
>marks the cline structs to say in which of them characters were
>missing.

I'm seeing a similar problem with the latest code.  I have a directory with
only three files named build.out{,1,2} and sometimes I see incorrect behavior
with completion.

If I try completing "bui", I sometimes get only "build" or more often I get the
following sequence from hitting tab:
   bui --> build.out[] --> build[.]out --> build.out[] --> build.out1[]
         --> build.out2[]
The cursor jumps back from the end to the dot, and then back to the end with
normal behavior.

-FR


__________________________________________________
Do You Yahoo!?
Send instant messages & get email alerts with Yahoo! Messenger.
http://im.yahoo.com/

PATCH: yet another match-spec-cursor bug

[Sven Wischnowsky]

With partial word matching and matches like `c6.yo' and `c6.yo.old',
the cursor was left on the dot. Caused by a thinko in the code that
marks the cline structs to say in which of them characters were
missing.


Bye
 Sven

Index: Src/Zle/compmatch.c
===================================================================
RCS file: /cvsroot/zsh/zsh/Src/Zle/compmatch.c,v
retrieving revision 1.9
diff -u -r1.9 compmatch.c
--- Src/Zle/compmatch.c	2000/04/26 06:54:27	1.9
+++ Src/Zle/compmatch.c	2000/04/28 12:38:55
@@ -1917,14 +1917,8 @@
 
 		if (tn) {
 		    diff = sub_join(o, n, tn, 0);
-
-		    if (po && po->prefix && cmp_anchors(n, pn, 0)) {
-			po->flags |= CLF_MISS;
-			po->max += diff;
-		    } else {
-			o->flags |= CLF_MISS;
-			o->max += diff;
-		    }
+		    o->flags |= CLF_MISS;
+		    o->max += diff;
 		    n = tn;
 		    po = o;
 		    o = o->next;

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