Re: compctl bug in 2.6-betat17

Re: compctl bug in 2.6-betat17

[Zoltan Hidvegi]

[-- Attachment #1: Type: application/pgp, Size: 3053 bytes --]

Re: bugs found

[Geoff Wing]

I wrote:
::From: hoh@approve.se (Goran Larsson)
::2)
::	$ PROMPT=AAA
::	AAA_
::	AAAPROMPT=AA
::	AA_
::	AAPROMPT=A
::	 _
::	 APROMPT='$'
::	 APROMPT='$ '
::	$ _
::	Note that any one character prompt is replaced by a space.
:OK. I've found the problem (and another smaller one - the code gets to a 
:place I commented as 'shouldn't happen' - now I've got to remember why :-)).

I am still not totally sure why I had it commented as `shouldn't happen'.
Most likely it would have had to do with cleareol stuff and it may be
redundant, but since it may not I've left it in.

:Patch in a day or so.

Or a bit longer.


*** zle_refresh.c.old	Wed Sep  4 02:12:23 1996
--- zle_refresh.c	Tue Sep 10 20:17:00 1996
***************
*** 515,521 ****
  	if (ln && nbuf[ln])
  	    strncpy(nl, p1, winw + 1);	/* next time obuf will be up-to-date */
  	else
! 	    nl = p1;		/* shouldn't happen */
  	nllen = winw;
      } else if (ollen > nllen) { /* make new line at least as long as old */
  	p1 = halloc(ollen + 1);
--- 515,521 ----
  	if (ln && nbuf[ln])
  	    strncpy(nl, p1, winw + 1);	/* next time obuf will be up-to-date */
  	else
! 	    nl = p1;		/* don't keep padding for prompt line */
  	nllen = winw;
      } else if (ollen > nllen) { /* make new line at least as long as old */
  	p1 = halloc(ollen + 1);
***************
*** 553,558 ****
--- 553,567 ----
  	if (*ol)
  	    ol++;
  	ccs = 1;
+     }
+ 
+ /* 2c: if we're on the first line, start checking at the end of the prompt;
+    we shouldn't be doing anything within the prompt */
+ 
+     if (ln == 0 && pptw) {
+ 	nl += (pptw - ccs);
+ 	ol += (pptw - ccs);
+ 	ccs = pptw;
      }
  
  /* 3: main display loop - write out the buffer using whatever tricks we can */
--
Geoff Wing [mason@primenet.com.au]   PrimeNet - Internet Consultancy
  Web: http://www.primenet.com.au/   Facsimile: +61-3-9819 3788

Re: 3.1.3 - config.status file generation problems...

[Andrew Main]

zefram wrote:
>It looks as if the shell you used to run the configure script has
>translated '\\' to '\' within single quotes, which is well out of order.
>Alternatively your ed has done the equivalent, which is also wrong.

On second thoughts, scratch that.  It's almost certainly echo that's
translating '\\' to '\'.  So the advice is, don't configure 3.1.3 with
a SysV sh.  zsh and bash will both work as intended (if invoked as sh).

Temporary fix: in the ed input in configure.in, change each '\' to '#',
and change ') | ed' to ') | sed s/\#/\\\\/g | ed'.  I'll produce a proper
patch when I have a chance to analyse it more thoroughly.

-zefram

Re: 3.1.3 - config.status file generation problems...

[Andrew J Cosgriff]

Bruce Stephens <b.stephens@isode.com> wrote:
>Andrew Main <zefram@tao.co.uk> writes:
>
>> As I suspected, you have a duff config.status, caused by bogus
>> editing.  It looks as if the shell you used to run the configure
>> script has translated '\\' to '\' within single quotes, which is
>> well out of order.
>
>Yes, that was it.  /bin/sh on Solaris 2.5.1 seems to do this (as does
>zsh-3.1.2-zefram3).  Explicitly using bash seems to have fixed it.


yup, same here :)


-- 
Andrew J Cosgriff <ajc@bing.wattle.id.au> exterminate all rational thought !
                       when asteroid landfill smile candy

Re: 3.1.3 - config.status file generation problems...

[Bruce Stephens]

Andrew Main <zefram@tao.co.uk> writes:

> As I suspected, you have a duff config.status, caused by bogus
> editing.  It looks as if the shell you used to run the configure
> script has translated '\\' to '\' within single quotes, which is
> well out of order.

Yes, that was it.  /bin/sh on Solaris 2.5.1 seems to do this (as does
zsh-3.1.2-zefram3).  Explicitly using bash seems to have fixed it.

Re: 3.1.3 - config.status file generation problems...

[Andrew J Cosgriff]

Andrew Main <zefram@tao.co.uk> wrote:
>Andrew J Cosgriff wrote:
>>2 goaway/2:a2 303 % ./config.status
>>creating Makefile
>>sed: Cannot find or open file :Makefile.in.
>
>As I suspected, you have a duff config.status, caused by bogus editing.
>It looks as if the shell you used to run the configure script has
>translated '\\' to '\' within single quotes, which is well out of order.
>Alternatively your ed has done the equivalent, which is also wrong.
>
>Please try replacing ed with GNU ed, so tht we can eliminate that
>possibility.  And what is your /bin/sh?

3 goaway/2:b9 43 % which ed
/usr/local/bin/ed

3 goaway/2:b9 44 % ed --version
GNU ed version 0.1

(looks like i was already using it...)

/bin/sh is the ordinary DU 4.0B /bin/sh

I'm just rerunning configure with zsh (3.0.5) - it'll take a little while to
run, so i'll report back a bit later with what happened.


-- 
Andrew J Cosgriff <ajc@bing.wattle.id.au> exterminate all rational thought !
                          air rhizome toxic boy blank

Re: 3.1.3 - config.status file generation problems...

[Bruce Stephens]

Andrew J Cosgriff <ajc@bing.wattle.id.au> writes:

> 2 goaway/2:a2 303 % ./config.status
> creating Makefile
> sed: Cannot find or open file :Makefile.in.
> creating Doc/Makefile
> sed: Cannot find or open file :Doc/Makefile.in.
> creating Etc/Makefile
> sed: Cannot find or open file :Etc/Makefile.in.
> creating Src/Makefile
> sed: Cannot find or open file :Src/Makefile.in.
> creating config.h
> cat: :config.h.in: No such file or directory
> config.h is unchanged
> 
> the sed in question is the ordinary one from DU 4.0B.
> 
> changing my path so it uses GNU sed 2.05 doesn't help any...
> 
> has this happened to anyone else ? the zefram4 release didn't have this
> problem...

Yes.  Exactly the same problem on sparc-sun-solaris2.5.1.  Haven't
looked at config.status, but configure itself fails in this way.

Re: 3.1.3 - config.status file generation problems...

[Andrew Main]

Andrew J Cosgriff wrote:
>2 goaway/2:a2 303 % ./config.status
>creating Makefile
>sed: Cannot find or open file :Makefile.in.

As I suspected, you have a duff config.status, caused by bogus editing.
It looks as if the shell you used to run the configure script has
translated '\\' to '\' within single quotes, which is well out of order.
Alternatively your ed has done the equivalent, which is also wrong.

Please try replacing ed with GNU ed, so tht we can eliminate that
possibility.  And what is your /bin/sh?

-zefram

Re: 3.1.3 - config.status file generation problems...

[Andrew Main]

Andrew J Cosgriff wrote:
>2 goaway/2:a2 303 % ./config.status
>creating Makefile
>sed: Cannot find or open file :Makefile.in.

Please send me your config.status.  It's possible that your ed behaves
differently from mine (GNU ed 0.2) and has mangled it.

>has this happened to anyone else ? the zefram4 release didn't have this
>problem...

zefram4 wasn't editing config.status after autoconf has had a go at it.

-zefram

3.1.3 - config.status file generation problems...

[Andrew J Cosgriff]

building on alpha-dec-osf4.0 with gcc 2.8.1 (not that it gets that far :)

when configure goes to create the files :

2 goaway/2:a2 303 % ./config.status
creating Makefile
sed: Cannot find or open file :Makefile.in.
creating Doc/Makefile
sed: Cannot find or open file :Doc/Makefile.in.
creating Etc/Makefile
sed: Cannot find or open file :Etc/Makefile.in.
creating Src/Makefile
sed: Cannot find or open file :Src/Makefile.in.
creating config.h
cat: :config.h.in: No such file or directory
config.h is unchanged

the sed in question is the ordinary one from DU 4.0B.

changing my path so it uses GNU sed 2.05 doesn't help any...

has this happened to anyone else ? the zefram4 release didn't have this
problem...

Thanks,
 Andrew
-- 
Andrew J Cosgriff <ajc@bing.wattle.id.au> exterminate all rational thought !
                    disposable guns reconstruction hope trip