problems re: makeinfo '--split-size' arg?

problems re: makeinfo '--split-size' arg?

[cgd]

So, i just tried building the head of the recently-created binutils
release branch (for target mips-elf, host x86-linux), and ran into:

/path/.../makeinfo: unrecognized option `--split-size=5000000'


My makeinfo has version:

> makeinfo (GNU texinfo) 4.0
> 
> Copyright (C) 1999 Free Software Foundation, Inc.
> There is NO warranty.  You may redistribute this software
> under the terms of the GNU General Public License.
> For more information about these matters, see the files named COPYING.

And the toplevel 'Makefile.tpl' says:

# For an installed makeinfo, we require it to be from texinfo 4 or
# higher, else we use the "missing" dummy.


While I realize that i should probably be using a newer 'makeinfo', I
cannot easily do so at this time.

I suspect that the top-level Makefile should be modified so that
it uses 'missing' instead of makeinfo if the --split-size option isn't
supported.  (Looks like it was added in rev 4.1 of the texinfo
package.)


I suspect that a diff like the one below is in order.  I tried it, and
it seems to DTRT for the version of makeinfo that i have (which is all
i tried 8-).

If approved, could somebody else please check it in for me?  (I hacked
it into my Makefile.in by hand to test it, but i don't have autogen
and wouldn't want to check in an improperly-generated file, etc...)


thanks,

chris
--
2003-04-28  Chris Demetriou  <cgd@broadcom.com>

	* Makefile.tpl (MAKEINFO): Require makeinfo 4.1 and later.
        * Makefile.in: Regenerate.

Index: Makefile.tpl
===================================================================
RCS file: /cvs/src/src/Makefile.tpl,v
retrieving revision 1.56.4.1
diff -u -p -r1.56.4.1 Makefile.tpl
--- Makefile.tpl	28 Apr 2003 02:25:36 -0000	1.56.4.1
+++ Makefile.tpl	28 Apr 2003 17:49:39 -0000
@@ -194,13 +194,13 @@ M4 = `if [ -f $$r/m4/m4 ] ; \
 	then echo $$r/m4/m4 ; \
 	else echo ${DEFAULT_M4} ; fi`
 
-# For an installed makeinfo, we require it to be from texinfo 4 or
+# For an installed makeinfo, we require it to be from texinfo 4.1 or
 # higher, else we use the "missing" dummy.
 MAKEINFO=@MAKEINFO@
 USUAL_MAKEINFO = `if [ -f $$r/texinfo/makeinfo/makeinfo ] ; \
 	then echo $$r/texinfo/makeinfo/makeinfo ; \
 	else if (makeinfo --version \
-	  | egrep 'texinfo[^0-9]*([1-3][0-9]|[4-9])') >/dev/null 2>&1; \
+	  | egrep 'texinfo[^0-9]*([1-3][0-9]|4\.[1-9]|[5-9])') >/dev/null 2>&1; \
         then echo makeinfo; else echo $$s/missing makeinfo; fi; fi`
 
 # This just becomes part of the MAKEINFO definition passed down to

Re: problems re: makeinfo '--split-size' arg?

[Joseph S. Myers]

On Mon, 28 Apr 2003 cgd@broadcom.com wrote:

> While I realize that i should probably be using a newer 'makeinfo', I
> cannot easily do so at this time.

And the binutils manuals "should" be using the more recently added Texinfo 
@copying feature (see the NEWS file in Texinfo 4.2 or later which says you 
should switch to it "as soon as possible (as in your next release)" to get 
proper copyright notices in HTML versions of the manuals).

> I suspect that the top-level Makefile should be modified so that
> it uses 'missing' instead of makeinfo if the --split-size option isn't
> supported.  (Looks like it was added in rev 4.1 of the texinfo
> package.)

For GCC it would be convenient to move the configure check for the
required Texinfo version (currently >= 4.2) to toplevel so features from
the same version of Texinfo can freely be used in all manuals, not just
those in the gcc/ subdirectory.  But other projects might not want the
required version to change whenever it's convenient to use some new
feature in some GCC manual and so the version needed for GCC is increased.

-- 
Joseph S. Myers
jsm28@cam.ac.uk

Re: problems re: makeinfo '--split-size' arg?

[Gerald Pfeifer]

On Mon, 28 Apr 2003 cgd@broadcom.com wrote:
>> makeinfo (GNU texinfo) 4.0

As Joseph (who is the expert on that) wrote we actually require makeinfo
4.2 for GCC and I agree that it should we checked automatically, as per
your patch.

Still, what I can do, and have in fact considered for the 3.3 branch also
before, is using --no-split instead of --split-size, as it shouldn't make
a difference there (except, in fact working around a "feature" of
--split-size in currently released versions of texinfo).

Bootstrapped and installed the 3.3-branch on i386-unknown-freebsd4.8,
committed to the 3.3-branch.

Gerald

2003-04-29  Gerald Pfeifer  <pfeifer@dbai.tuwien.ac.at>

	* Makefile.tpl (MAKEINFOFLAGS): Use --no-split instead of
	--split-size.
	* Makefile.in: Regenerate.

Index: Makefile.in
===================================================================
RCS file: /cvs/gcc/gcc/Makefile.in,v
retrieving revision 1.129.2.5
diff -u -3 -p -r1.129.2.5 Makefile.in
--- Makefile.in	25 Apr 2003 20:27:39 -0000	1.129.2.5
+++ Makefile.in	29 Apr 2003 14:04:18 -0000
@@ -163,7 +163,7 @@ MAKEINFO = `if [ -f $$r/texinfo/makeinfo
 # sub-makes.  It lets flags be given on the command line while still
 # using the makeinfo from the object tree.
 # (Default to avoid splitting info files by setting the threshold high.)
-MAKEINFOFLAGS = --split-size=5000000
+MAKEINFOFLAGS = --no-split

 EXPECT = `if [ -f $$r/expect/expect ] ; \
 	then echo $$r/expect/expect ; \
Index: Makefile.tpl
===================================================================
RCS file: /cvs/gcc/gcc/Makefile.tpl,v
retrieving revision 1.13.6.4
diff -u -3 -p -r1.13.6.4 Makefile.tpl
--- Makefile.tpl	25 Apr 2003 20:27:39 -0000	1.13.6.4
+++ Makefile.tpl	29 Apr 2003 14:04:19 -0000
@@ -166,7 +166,7 @@ MAKEINFO = `if [ -f $$r/texinfo/makeinfo
 # sub-makes.  It lets flags be given on the command line while still
 # using the makeinfo from the object tree.
 # (Default to avoid splitting info files by setting the threshold high.)
-MAKEINFOFLAGS = --split-size=5000000
+MAKEINFOFLAGS = --no-split

 EXPECT = `if [ -f $$r/expect/expect ] ; \
 	then echo $$r/expect/expect ; \

Re: problems re: makeinfo '--split-size' arg?

[Steven Bosscher]

Op di 29-04-2003, om 16:06 schreef Gerald Pfeifer:
> Index: Makefile.in
> ===================================================================
> RCS file: /cvs/gcc/gcc/Makefile.in,v
> retrieving revision 1.129.2.5
> diff -u -3 -p -r1.129.2.5 Makefile.in
> --- Makefile.in	25 Apr 2003 20:27:39 -0000	1.129.2.5
> +++ Makefile.in	29 Apr 2003 14:04:18 -0000
> @@ -163,7 +163,7 @@ MAKEINFO = `if [ -f $$r/texinfo/makeinfo
>  # sub-makes.  It lets flags be given on the command line while still
>  # using the makeinfo from the object tree.
>  # (Default to avoid splitting info files by setting the threshold high.)
> -MAKEINFOFLAGS = --split-size=5000000
> +MAKEINFOFLAGS = --no-split

I suppose the comment should be changed here, then?  We don't set a
threshold, so the comment doesn't make sense anymore.

> 
>  EXPECT = `if [ -f $$r/expect/expect ] ; \
>  	then echo $$r/expect/expect ; \
> Index: Makefile.tpl
> ===================================================================
> RCS file: /cvs/gcc/gcc/Makefile.tpl,v
> retrieving revision 1.13.6.4
> diff -u -3 -p -r1.13.6.4 Makefile.tpl
> --- Makefile.tpl	25 Apr 2003 20:27:39 -0000	1.13.6.4
> +++ Makefile.tpl	29 Apr 2003 14:04:19 -0000
> @@ -166,7 +166,7 @@ MAKEINFO = `if [ -f $$r/texinfo/makeinfo
>  # sub-makes.  It lets flags be given on the command line while still
>  # using the makeinfo from the object tree.
>  # (Default to avoid splitting info files by setting the threshold high.)
> -MAKEINFOFLAGS = --split-size=5000000
> +MAKEINFOFLAGS = --no-split

Same here.

> 
>  EXPECT = `if [ -f $$r/expect/expect ] ; \
>  	then echo $$r/expect/expect ; \
> 


Greetz
Steven

Re: problems re: makeinfo '--split-size' arg?

[Gerald Pfeifer]

On Tue, 29 Apr 2003, Steven Bosscher wrote:
> I suppose the comment should be changed here, then?  We don't set a
> threshold, so the comment doesn't make sense anymore.

Fixed.

Gerald
-- 
Gerald "Jerry"   pfeifer@dbai.tuwien.ac.at   http://www.pfeifer.com/gerald/

Re: problems re: makeinfo '--split-size' arg?

[cgd]

At Mon, 28 Apr 2003 18:56:56 +0000 (UTC), "Joseph S. Myers" wrote:
> > I suspect that the top-level Makefile should be modified so that
> > it uses 'missing' instead of makeinfo if the --split-size option isn't
> > supported.  (Looks like it was added in rev 4.1 of the texinfo
> > package.)
> 
> For GCC it would be convenient to move the configure check for the
> required Texinfo version (currently >= 4.2) to toplevel so features from
> the same version of Texinfo can freely be used in all manuals, not just
> those in the gcc/ subdirectory.  But other projects might not want the
> required version to change whenever it's convenient to use some new
> feature in some GCC manual and so the version needed for GCC is increased.

It doesn't seem nice to mandate a frequently-increasing version being
required across the whole source tree, unless that version is *really*
necessary across the source tree programs.

On the other hand, there's something to be said for rebuilding info
files consistently across the source tree (i.e., if they're not
rebuilt in one place, they're not rebuilt in all).

If people think that a patch like mine, but w/ the version bumped to
4.2, is reasonable, then that's fine w/ me...  All i know is that the
current state allows one to use makeinfo 4.0, which fails during the
build.  8-)



cgd

Re: problems re: makeinfo '--split-size' arg?

[Gerald Pfeifer]

On Thu, 1 May 2003 cgd@broadcom.com wrote:
> If people think that a patch like mine, but w/ the version bumped to
> 4.2, is reasonable, then that's fine w/ me...  All i know is that the
> current state allows one to use makeinfo 4.0, which fails during the
> build.  8-)

That's certainly a bug, agreed.  Would you mind submitting an updated
version of your patch so that the GCC Makefile/configury maintainers
can have a look?

Thanks,
Gerald
-- 
Gerald "Jerry"   pfeifer@dbai.tuwien.ac.at   http://www.pfeifer.com/gerald/

[patch]

[cgd]

At Fri, 2 May 2003 08:23:00 +0200 (CEST), Gerald Pfeifer wrote:
> Would you mind submitting an updated
> version of your patch so that the GCC Makefile/configury maintainers
> can have a look?

It's below.  And this time i've sent it to the right place.  (my
original msg wasn't going to include a patch, and i forgot to change
gcc -> gcc-patches when i added the patch.  8-)

I don't have autogen, so i just applied the same patch to makefile.in,
and tested the result w/ texinfo 4.0 to build mipsisa64-elf.  I've not
tested this w/ a 'good' version of texinfo / makeinfo.

If approved, I'd appreciate it if somebody else would apply this
(since i'd rather have them actually regenerate Makefile.in than have
me hack it.  8-)


cgd
--
2003-05-02  Chris Demetriou  <cgd@broadcom.com>

	* Makefile.tpl: Require "makeinfo" from texinfo 4.2 or later.
	* Makefile.in: Regenerate.

Index: Makefile.tpl
===================================================================
RCS file: /cvs/gcc/gcc/Makefile.tpl,v
retrieving revision 1.44
diff -u -p -r1.44 Makefile.tpl
--- Makefile.tpl	28 Apr 2003 02:23:46 -0000	1.44
+++ Makefile.tpl	2 May 2003 21:10:25 -0000
@@ -194,13 +194,13 @@ M4 = `if [ -f $$r/m4/m4 ] ; \
 	then echo $$r/m4/m4 ; \
 	else echo ${DEFAULT_M4} ; fi`
 
-# For an installed makeinfo, we require it to be from texinfo 4 or
+# For an installed makeinfo, we require it to be from texinfo 4.2 or
 # higher, else we use the "missing" dummy.
 MAKEINFO=@MAKEINFO@
 USUAL_MAKEINFO = `if [ -f $$r/texinfo/makeinfo/makeinfo ] ; \
 	then echo $$r/texinfo/makeinfo/makeinfo ; \
 	else if (makeinfo --version \
-	  | egrep 'texinfo[^0-9]*([1-3][0-9]|[4-9])') >/dev/null 2>&1; \
+	  | egrep 'texinfo[^0-9]*([1-3][0-9]|4\.[2-9]|[5-9])') >/dev/null 2>&1; \
         then echo makeinfo; else echo $$s/missing makeinfo; fi; fi`
 
 # This just becomes part of the MAKEINFO definition passed down to

[patch] require makeinfo 4.2 or better

[cgd]

At Fri, 2 May 2003 08:23:00 +0200 (CEST), Gerald Pfeifer wrote:
> Would you mind submitting an updated
> version of your patch so that the GCC Makefile/configury maintainers
> can have a look?

It's below.  And this time i've sent it to the right place.  (my
original msg wasn't going to include a patch, and i forgot to change
gcc -> gcc-patches when i added the patch.  8-)

I don't have autogen, so i just applied the same patch to makefile.in,
and tested the result w/ texinfo 4.0 to build mipsisa64-elf.  I've not
tested this w/ a 'good' version of texinfo / makeinfo.

If approved, I'd appreciate it if somebody else would apply this
(since i'd rather have them actually regenerate Makefile.in than have
me hack it.  8-)


cgd
--
2003-05-02  Chris Demetriou  <cgd@broadcom.com>

	* Makefile.tpl: Require "makeinfo" from texinfo 4.2 or later.
	* Makefile.in: Regenerate.

Index: Makefile.tpl
===================================================================
RCS file: /cvs/gcc/gcc/Makefile.tpl,v
retrieving revision 1.44
diff -u -p -r1.44 Makefile.tpl
--- Makefile.tpl	28 Apr 2003 02:23:46 -0000	1.44
+++ Makefile.tpl	2 May 2003 21:10:25 -0000
@@ -194,13 +194,13 @@ M4 = `if [ -f $$r/m4/m4 ] ; \
 	then echo $$r/m4/m4 ; \
 	else echo ${DEFAULT_M4} ; fi`
 
-# For an installed makeinfo, we require it to be from texinfo 4 or
+# For an installed makeinfo, we require it to be from texinfo 4.2 or
 # higher, else we use the "missing" dummy.
 MAKEINFO=@MAKEINFO@
 USUAL_MAKEINFO = `if [ -f $$r/texinfo/makeinfo/makeinfo ] ; \
 	then echo $$r/texinfo/makeinfo/makeinfo ; \
 	else if (makeinfo --version \
-	  | egrep 'texinfo[^0-9]*([1-3][0-9]|[4-9])') >/dev/null 2>&1; \
+	  | egrep 'texinfo[^0-9]*([1-3][0-9]|4\.[2-9]|[5-9])') >/dev/null 2>&1; \
         then echo makeinfo; else echo $$s/missing makeinfo; fi; fi`
 
 # This just becomes part of the MAKEINFO definition passed down to