Another patch in need of explanatory context ...

This patch is starting to smell like a fix for the several
subtle I18N problems on HEAD. The segfault, and
the assertion failure all have RPM_I18NSTRING_TYPE
in common when I drill down to see what is happening.

(aside) I have exactly 1 tag in 1 package that uses
RPM_I18NSTRING_TYPE, and the only reason I
have even that is due to a Fedora packaging
oversight. 1 tag is enough for a reproducer, but hardly
enough to claim "fixed". Off to build some pkgs
with i18n tags this evening, todo++.

This patch changes how RPM_I18NSTRING_TYPE
is handled. The patch treats i18n tags as an argv array
rather than a string (which is how i18n strings
are handled internal to rpm).

I'm going to try to hustle a fix into HEAD
to stop creating packages that use RPM_I18NSTRING_TYPE.
For legacy reasons, rpm-5.0 is going to have to continue
handling the bad old type. What I need to find out
is whether arbitrary tags behave well enough to risk throwing
Summary: Group: and Description: onto the mechanism.
So far everything I've tried with arbitrary tags has worked, but
those thre tags are a bit more important than most info in
packages.

73 de Jeff

On Dec 12, 2007, at 4:55 PM, Jeff Johnson wrote:

>   RPM Package Manager, CVS Repository
>   http://rpm5.org/cvs/
>    
> ______________________________________________________________________ 
> ______
>
>   Server: rpm5.org                         Name:   Jeff Johnson
>   Root:   /v/rpm/cvs                       Email:  jbj@rpm5.org
>   Module: rpm                              Date:   12-Dec-2007  
> 22:55:26
>   Branch: HEAD                             Handle: 2007121221552500
>
>   Modified files:
>     rpm                     CHANGES
>     rpm/rpmdb               header.c
>
>   Log:
>     - another attempt to identify the RPM_I18NSTRING_TYPE flaw.
>
>   Summary:
>     Revision    Changes     Path
>     1.1979      +1  -0      rpm/CHANGES
>     1.144       +1  -11     rpm/rpmdb/header.c
>    
> ______________________________________________________________________ 
> ______
>
>   patch -p0 <<'@@ .'
>   Index: rpm/CHANGES
>    
> ====================================================================== 
> ======
>   $ cvs diff -u -r1.1978 -r1.1979 CHANGES
>   --- rpm/CHANGES	12 Dec 2007 20:01:24 -0000	1.1978
>   +++ rpm/CHANGES	12 Dec 2007 21:55:25 -0000	1.1979
>   @@ -1,4 +1,5 @@
>    5.0b1 -> 5.0b2:
>   +    - jbj: another attempt to identify the RPM_I18NSTRING_TYPE  
> flaw.
>        - jbj: attempt to measure performance improvement using  
> binary tag search.
>        - jbj: trade an I18NSTRING assert failure for a memory leak  
> for the moment.
>        - jbj: push rpm2cpio w-i-p into cvs, try to measure linear <- 
> > bsearch win.
>   @@ .
>   patch -p0 <<'@@ .'
>   Index: rpm/rpmdb/header.c
>    
> ====================================================================== 
> ======
>   $ cvs diff -u -r1.143 -r1.144 header.c
>   --- rpm/rpmdb/header.c	12 Dec 2007 20:01:24 -0000	1.143
>   +++ rpm/rpmdb/header.c	12 Dec 2007 21:55:25 -0000	1.144
>   @@ -1893,23 +1893,13 @@
>        case RPM_UINT64_TYPE:
>    	nb = he->c * sizeof(*he->p.ui64p);
>    	break;
>   -    case RPM_I18NSTRING_TYPE:
>   -#if HACK_AROUND
>   -assert(he->c == 1);	/* XXX stop unimplemented oversights. */
>   -#else
>   -	if (he->c > 1) {
>   -	    he->t = RPM_STRING_TYPE;
>   -	    he->p.str = he->p.argv[0];
>   -	    he->c = 1;
>   -	}
>   -#endif
>   -	/*@fallthrough@*/
>        case RPM_STRING_TYPE:
>    	if (he->p.str)
>    	    nb = strlen(he->p.str) + 1;
>    	else
>    	    rc = 0;
>    	break;
>   +    case RPM_I18NSTRING_TYPE:
>        case RPM_STRING_ARRAY_TYPE:
>    	break;
>        }
>   @@ .
> ______________________________________________________________________
> RPM Package Manager                                    http://rpm5.org
> CVS Sources Repository                                rpm-cvs@rpm5.org