On Wed, Mar 31, 2010 at 03:59:44AM +0600, Alexandr Shadchin wrote:
> Hi, Ports!
>
> This update package SBCL to the latest release 1.0.37.
Is there any particular reason you need this version of SBCL, or is it
just because it's several months old by now?
> Tested on i386.
I assume this means that 'make regress' showed only expected failures.
> Comments ? OK ?
I don't see any need for this update unless it actually provides
some real advantage over the existing version. However I have no
objection as long as the regress tests pass on both i386 and amd64.
> also need to patch texinfo
>
> makeinfo -html --css-include=foo.css bar.texinfo
>
> Opens file foo.css, but not close it, which may lead to error:
> Too many open files
>
> I received this error when porting sbcl-1.0.37
Hm, interesting. Are you saying that this one extra open file causes
makeinfo to hit the fd limit? Or does it open the file in a loop
without closing it? Perhaps the manual has simply grown complex enough
that you need higher resource limits to build it.
> --
> Alexandr Shadchin
Comments on the diff follow.
> Index: Makefile
> ===================================================================
> RCS file: /cvs/ports/lang/sbcl/Makefile,v
> retrieving revision 1.3
> diff -u -p -r1.3 Makefile
> --- Makefile 17 Nov 2009 10:44:59 -0000 1.3
> +++ Makefile 30 Mar 2010 21:50:04 -0000
> @@ -1,11 +1,11 @@
> # $OpenBSD: Makefile,v 1.3 2009/11/17 10:44:59 pirofti Exp $
>
> # not yet ported to other arches
> -ONLY_FOR_ARCHS = amd64 i386
> +ONLY_FOR_ARCHS= amd64 i386
What is the reason for all these whitespace changes? I believe a space
after the variable name is the preferred style now.
[snip]
> Index: patches/patch-src_runtime_bsd-os_c
> ===================================================================
> RCS file: /cvs/ports/lang/sbcl/patches/patch-src_runtime_bsd-os_c,v
> retrieving revision 1.1
> diff -u -p -r1.1 patch-src_runtime_bsd-os_c
> --- patches/patch-src_runtime_bsd-os_c 17 Nov 2009 10:45:00 -0000
> 1.1
> +++ patches/patch-src_runtime_bsd-os_c 30 Mar 2010 21:50:04 -0000
> @@ -4,9 +4,9 @@ Add a flag to os_get_runtime_executable_
> externally usable path is required, which means we can't use
> /proc/curproc/file even if procfs happens to be mounted.
>
> ---- src/runtime/bsd-os.c.orig Wed May 13 11:13:48 2009
> -+++ src/runtime/bsd-os.c Mon Jun 29 07:53:42 2009
> -@@ -465,16 +465,12 @@ os_get_runtime_executable_path()
> +--- src/runtime/bsd-os.c.orig Sun Oct 25 13:25:01 2009
> ++++ src/runtime/bsd-os.c Fri Mar 5 23:26:24 2010
> +@@ -456,16 +456,12 @@ os_get_runtime_executable_path()
> }
> #elif defined(LISP_FEATURE_NETBSD) || defined(LISP_FEATURE_OPENBSD)
> char *
Perhaps someone with more porting experience can weigh in on this. Is
it preferred to update a diff if nothing changed other than line
numbers and dates, but the diff still applies fine?
[snip]
> Index: makeinfo/html.c
> ===================================================================
> RCS file: /cvs/src/gnu/usr.bin/texinfo/makeinfo/html.c,v
> retrieving revision 1.2
> diff -u -p -r1.2 html.c
> --- makeinfo/html.c 17 Jul 2006 22:29:29 -0000 1.2
> +++ makeinfo/html.c 6 Mar 2010 11:27:38 -0000
> @@ -167,6 +167,9 @@ process_css_file (char *filename)
> lastchar = c;
> }
>
> + if (f != stdin)
> + fclose(f);
> +
> /* Reached the end of the file. We should not be still in a comment. */
> if (state == comment_state)
> warning (_("%s:%d: --css-file ended in comment"), filename, lineno);
You should probably submit this separately with sendbug.
