Re: refcard.dvi build patch

Re: refcard.dvi build patch

[Colin Walters]

Andrew Cagney <ac131313@cygnus.com> writes:

> Can anyone think of a reason to not set REFEDITS to /dev/null by
> default?  Would either cygwin or djgpp have problems?

No idea there, but /dev/null is definitely a better idea if it works.

> PS: Which version of bash?

I'm using the version from Debian sid (2.05-4).

Re: refcard.dvi build patch

[Eli Zaretskii]

On Wed, 27 Jun 2001, Andrew Cagney wrote:

> > I needed the following patch to build refcard.dvi on my system; bash
> > was complaining about the for f in ; do ...; done syntax when
> > when $REFEDITS was empty.
> 
> Can anyone think of a reason to not set REFEDITS to /dev/null by 
> default?  Would either cygwin or djgpp have problems?

I don't expect them to have problems (/dev/null is supported), but
please post the patch and I will try it.

Btw, we had this problem pop up before, and IIRC other shells, besides
Bash, had problems with some of the suggested solutions.  It would be
nice to make sure the proposed one doesn't break them.

I don't remember the details, but they can be traced in the mailing
list archives.

Re: refcard.dvi build patch

[Colin Walters]

Eli Zaretskii <eliz@is.elta.co.il> writes:

> I don't expect them to have problems (/dev/null is supported), but
> please post the patch and I will try it.

I posted it to gdb-patches; do you want me to send another copy to
you?  It's also available here:

<URL: http://sources.redhat.com/ml/gdb-patches/2001-06/msg00370.html >

> Btw, we had this problem pop up before, and IIRC other shells,
> besides Bash, had problems with some of the suggested solutions.  It
> would be nice to make sure the proposed one doesn't break them.

I deliberately didn't add any new shell code, so it shouldn't fail
anywhere the previous one didn't.

Re: refcard.dvi build patch

[Eli Zaretskii]

> From: Colin Walters <walters@cis.ohio-state.edu>
> Date: Fri, 29 Jun 2001 02:06:22 -0400
> 
> Eli Zaretskii <eliz@is.elta.co.il> writes:
> 
> > I don't expect them to have problems (/dev/null is supported), but
> > please post the patch and I will try it.
> 
> I posted it to gdb-patches; do you want me to send another copy to
> you?

I meant the patch ANdrew was thinking about, the one with /dev/null.

> > Btw, we had this problem pop up before, and IIRC other shells,
> > besides Bash, had problems with some of the suggested solutions.  It
> > would be nice to make sure the proposed one doesn't break them.
> 
> I deliberately didn't add any new shell code, so it shouldn't fail
> anywhere the previous one didn't.

But that's just it: the previous code did fail with some shells.

Re: refcard.dvi build patch

[Andrew Cagney]

ChrisF?  Ok in theory on cygwin?

	Andrew

> I needed the following patch to build refcard.dvi on my system; bash
> was complaining about the for f in ; do ...; done syntax when
> when $REFEDITS was empty.
> 
> 
> Can anyone think of a reason to not set REFEDITS to /dev/null by default?  Would either cygwin or djgpp have problems?
> 
>     Andrew
> 
> PS: Which version of bash?
> 
> 
> 
> 2001-06-21  Colin Walters  <walters@cis.ohio-state.edu>
> 
> * identity.sed: New file.
>     * Makefile.in (REFEDITS): Use it.
>     (refcard.dvi): Remove logic for testing for empty REFEDITS; this
>     is now handled by identity.sed instead.
> 
> --- /dev/null    Wed Dec 31 19:00:00 1969
> +++ identity.sed    Thu Jun 21 15:55:31 2001
> @@ -0,0 +1 @@
> +# This is an empty file used by the Makefile for building refcard.dvi.
> 
> --- Makefile.in.~1.13.~    Mon May 14 12:18:50 2001
> +++ Makefile.in    Thu Jun 21 13:31:02 2001
> @@ -84,8 +84,8 @@
>  # e.g. for A4, Postscript:  REFEDITS = a4rc.sed psrc.sed
>  #      for A4, CM fonts:    REFEDITS = a4rc.sed
>  #      for US, PS fonts:    REFEDITS = psrc.sed
> -#      for default:
> -REFEDITS =
> +#      for default:         REFEDITS = identity.sed
> +REFEDITS = identity.sed
>  
>  # Don Knuth's TeX formatter
>  TEX = tex
> @@ -200,14 +200,10 @@
>  
>  # GDB QUICK REFERENCE (dvi output)
>  refcard.dvi : refcard.tex $(REFEDITS)
> -    if [ -z "$(REFEDITS)" ]; then \
> -        cp $(srcdir)/refcard.tex sedref.tex ; \
> -    else \
> -        echo > tmp.sed ; \
> -        for f in $(REFEDITS) ; do \
> -            cat $(srcdir)/$$f >>tmp.sed ; done ; \
> -        sed -f tmp.sed $(srcdir)/refcard.tex >sedref.tex ; \
> -    fi
> +    echo > tmp.sed ; \
> +    for f in $(REFEDITS) ; do \
> +        cat $(srcdir)/$$f >>tmp.sed ; done ; \
> +    sed -f tmp.sed $(srcdir)/refcard.tex >sedref.tex ; \
>      $(SET_TEXINPUTS) $(TEX) sedref.tex
>      mv sedref.dvi refcard.dvi
>      rm -f sedref.log sedref.tex tmp.sed
> 
> 
> 
> 
> 

Re: refcard.dvi build patch

[Andrew Cagney]

> I needed the following patch to build refcard.dvi on my system; bash
> was complaining about the for f in ; do ...; done syntax when
> when $REFEDITS was empty.
> 
> 
> Can anyone think of a reason to not set REFEDITS to /dev/null by default?

> Would either cygwin or djgpp have problems?


Hmm, I can.  It would really convolute:

        for f in $(REFEDITS) ; do \
            cat $(srcdir)/$$f >>tmp.sed ; done ; \

since ``/dev/null'' would result in ``$(srcdir)//dev/null'' :-/  If no 
one can see problems with the original patch I think it or:

	for f in x $(REFEDITS) ; do \
	  test x$$f = xx && continue ; \
	  cat $(srcdir)/$$f >> tmp.sed ; \
	done

are better.

Eli, preference?  I'll cook up the patch.

	Andrew

PS:

 > +    echo > tmp.sed ; \
 > +    for f in $(REFEDITS) ; do \
 > +        cat $(srcdir)/$$f >>tmp.sed ; done ; \
 > +    sed -f tmp.sed $(srcdir)/refcard.tex >sedref.tex ; \


can be simplified further.  Most of the trailing ``\'' have also been 
made redundant.



>     Andrew
> 
> PS: Which version of bash?
> 
> 
> 
> 2001-06-21  Colin Walters  <walters@cis.ohio-state.edu>
> 
> * identity.sed: New file.
>     * Makefile.in (REFEDITS): Use it.
>     (refcard.dvi): Remove logic for testing for empty REFEDITS; this
>     is now handled by identity.sed instead.
> 
> --- /dev/null    Wed Dec 31 19:00:00 1969
> +++ identity.sed    Thu Jun 21 15:55:31 2001
> @@ -0,0 +1 @@
> +# This is an empty file used by the Makefile for building refcard.dvi.
> 
> --- Makefile.in.~1.13.~    Mon May 14 12:18:50 2001
> +++ Makefile.in    Thu Jun 21 13:31:02 2001
> @@ -84,8 +84,8 @@
>  # e.g. for A4, Postscript:  REFEDITS = a4rc.sed psrc.sed
>  #      for A4, CM fonts:    REFEDITS = a4rc.sed
>  #      for US, PS fonts:    REFEDITS = psrc.sed
> -#      for default:
> -REFEDITS =
> +#      for default:         REFEDITS = identity.sed
> +REFEDITS = identity.sed
>  
>  # Don Knuth's TeX formatter
>  TEX = tex
> @@ -200,14 +200,10 @@
>  
>  # GDB QUICK REFERENCE (dvi output)
>  refcard.dvi : refcard.tex $(REFEDITS)
> -    if [ -z "$(REFEDITS)" ]; then \
> -        cp $(srcdir)/refcard.tex sedref.tex ; \
> -    else \
> -        echo > tmp.sed ; \
> -        for f in $(REFEDITS) ; do \
> -            cat $(srcdir)/$$f >>tmp.sed ; done ; \
> -        sed -f tmp.sed $(srcdir)/refcard.tex >sedref.tex ; \
> -    fi
> +    echo > tmp.sed ; \
> +    for f in $(REFEDITS) ; do \
> +        cat $(srcdir)/$$f >>tmp.sed ; done ; \
> +    sed -f tmp.sed $(srcdir)/refcard.tex >sedref.tex ; \
>      $(SET_TEXINPUTS) $(TEX) sedref.tex
>      mv sedref.dvi refcard.dvi
>      rm -f sedref.log sedref.tex tmp.sed
> 
> 
> 
> 
> 

Re: refcard.dvi build patch

[Eli Zaretskii]

On Wed, 4 Jul 2001, Andrew Cagney wrote:

> If no one can see problems with the original patch I think it or:
> 
> 	for f in x $(REFEDITS) ; do \
> 	  test x$$f = xx && continue ; \
> 	  cat $(srcdir)/$$f >> tmp.sed ; \
> 	done
> 
> are better.
> 
> Eli, preference?

This seems fine to me.

Re: refcard.dvi build patch

[Andrew Cagney]

Try the attached?

	Andrew
From ac131313@cygnus.com Thu Jul 05 22:22:00 2001
From: Andrew Cagney <ac131313@cygnus.com>
To: Daniel Jacobowitz <dmj+@andrew.cmu.edu>
Cc: gdb-patches@sources.redhat.com
Subject: Re: [RFA] mips32_next_pc overdue cleanup
Date: Thu, 05 Jul 2001 22:22:00 -0000
Message-id: <3B454B1B.7040506@cygnus.com>
References: <20010704114147.A15695@nevyn.them.org>
X-SW-Source: 2001-07/msg00140.html
Content-length: 823

> This patch is half style cleanups and half obvious fixes, but there's so
> many I'd appreciate more eyes.  According to the masks in the opcodes
> directory, these changes are rigt, and what we had before can't possibly
> work.  Especially the fact that we would never, ever predict a BNE as taken
> before these changes convinced me that they really were correct and this
> code had never been particularly tested.  Comments?


Approved.

History?  It pre-dates my MIPS hacking :-)  Dig dig, ah, it started out 
as part of some prototyping that wasn't finished  It was added in '98. 
The WinCE GDB port ended up using it so I suspect it was made to mostly 
work.  The main reason the code worked is because it didn't bother to 
use the broken macros that were added at the same time as the original 
code :-/

	Andrew

Re: refcard.dvi build patch

[Eli Zaretskii]

On Fri, 6 Jul 2001, Andrew Cagney wrote:

> Try the attached?

It works with the DJGPP port.

Re: refcard.dvi build patch

[Andrew Cagney]

> 2001-07-06  Andrew Cagney  <ac131313@redhat.com>
> 
> * Makefile.in (refcard.dvi): Rewrite to avoid problems with empty
>     `test` expressions on bash.  Problem reported by Colin Walters.

I've checked this in.

	Andrew

Re: refcard.dvi build patch

[Colin Walters]

Andrew Cagney <ac131313@cygnus.com> writes:

> Try the attached?
> 
> 	Andrew
> 2001-07-06  Andrew Cagney  <ac131313@redhat.com>
> 
> 	* Makefile.in (refcard.dvi): Rewrite to avoid problems with empty
> 	`test` expressions on bash.  Problem reported by Colin Walters.

Works for me.  Sorry about the delay in getting back to you.