CVS srivasta: Manoj
CVSROOT:/cvs/debian-policy Module name:debian-policy Changes by: srivastaThu Nov 14 23:44:23 MST 2002 Modified files: . : menu-policy.sgml policy.sgml upgrading-checklist.html debconf_spec : debconf_specification.xml debconf_spec/include: commands.xml debian : changelog Log message: Manoj * Added example for why one may call ldconfig anywhere in the postsint.closes: Bug#120585 * Add the modifications about base system, as opposed to the soon to be obsolete base section (I assume it is)closes: Bug#53582 * Rearranged the virtual packages list. closes: Bug#72980 * This is basically an attempt to ratify the current practice of using debhelper in the clean target. Currently, policy does not require debhelper to be installed when the clean target is run, even though it is in the build-depends field. This was a simple oversight. closes: Bug#164035 * No longer depend on fileutils. closes: Bug#167425 * Added the Apps/Science menu section closes: Bug#162812 * Applied text patch from Joey Hess to the debconf spec simply make it conform to the reality of how some things work now. This is part of an effort to make debconf and cdebconf better substitutes for each other. Since it was not an XML patch, no special markup is present in the new content, except where I made guesses.closes: Bug#160776 * Clarify section 13.3.closes: Bug#160248 * Removed the undocumented(7) hack requirement.closes: Bug#39830 Josip * Removed the obsolete notion of documenting changes within the copyright file, closes: Bug#65764
CVS srivasta: there is no id in a footnote?
CVSROOT:/cvs/debian-policy Module name:debian-policy Changes by: srivastaThu Nov 14 23:49:40 MST 2002 Modified files: . : policy.sgml Log message: there is no id in a footnote?
CVS srivasta: update standards version
CVSROOT:/cvs/debian-policy Module name:debian-policy Changes by: srivastaThu Nov 14 23:52:48 MST 2002 Modified files: debian : control Log message: update standards version
Re: Bug#167422: files in /usr/share should be world-readable
On Thu, Nov 14, 2002 at 06:13:50PM -0600, Manoj Srivastava wrote: Branden == Branden Robinson [EMAIL PROTECTED] writes: Branden Well, according to the FHS, you shouldn't be putting variabel data in Branden /usr at all: It is not variable data. It is static data; that does not change after installation. /usr/ is allowed to change when you install a new package. It still seems wrong to have log files in /usr. Are the files ever accessed by anything other than the user? I.e., do the install logs contain information that is useful to Emacs itself? -- G. Branden Robinson| There's nothing an agnostic can't Debian GNU/Linux | do if he doesn't know whether he [EMAIL PROTECTED] | believes in it or not. http://people.debian.org/~branden/ | -- Graham Chapman pgpiwsRmBwCMV.pgp Description: PGP signature
Re: Bug#167422: files in /usr/share should be world-readable
Branden == Branden Robinson [EMAIL PROTECTED] writes: Branden On Thu, Nov 14, 2002 at 06:13:50PM -0600, Manoj Srivastava wrote: Branden == Branden Robinson [EMAIL PROTECTED] writes: Branden Well, according to the FHS, you shouldn't be putting variabel data in Branden /usr at all: It is not variable data. It is static data; that does not change after installation. /usr/ is allowed to change when you install a new package. Branden It still seems wrong to have log files in /usr. If these were ordinary log files of a running program generating data, I may agree. But since we are down to gut feelings, a bunch of files are generated by postinst, and they all reside in the same place. Some of these files are read by emacs; some times there is a README read by the inquisitive users, and then there is a record of what transpired while gemerating these files. It strikes my gut as sane that these files, all generated by the postinst, live in the same location. Branden Are the files ever accessed by anything other than the user? Branden I.e., do the install logs contain information that is useful Branden to Emacs itself? They are not even accessed by the user directly, and certainly not by emacs. manoj -- Paranoid schizophrenics outnumber their enemies at least two to one. Manoj Srivastava [EMAIL PROTECTED] http://www.debian.org/%7Esrivasta/ 1024R/C7261095 print CB D9 F4 12 68 07 E4 05 CC 2D 27 12 1D F5 E8 6E 1024D/BF24424C print 4966 F272 D093 B493 410B 924B 21BA DABB BF24 424C
Bug#62768: And now for the 2 years and older (bugs closed as well)
Ian == Ian Zimmerman [EMAIL PROTECTED] writes: Ian The inconsistency makes it unnecessarily hard to write automated Ian scripts that build a kernel and all modules. Sometimes you need to do Ian the tar -x before running make-kpkg, sometimes not. Ian Not necessarily a policy issue, but definitely a bug. I tend to agree. manoj -- You love peace. Manoj Srivastava [EMAIL PROTECTED] http://www.debian.org/%7Esrivasta/ 1024R/C7261095 print CB D9 F4 12 68 07 E4 05 CC 2D 27 12 1D F5 E8 6E 1024D/BF24424C print 4966 F272 D093 B493 410B 924B 21BA DABB BF24 424C
Bug#165063: debian-policy: Section `12.8.3 Packages providing a terminal emulator' fails to sufficently document the -e option
I suggest that the following patch be applied to policy in order to resolve this issue: --- policy.sgml.old Fri Nov 15 09:30:47 2002 +++ policy.sgml Fri Nov 15 09:36:16 2002 @@ -6982,9 +6982,12 @@ emulator application were so coded, be a new view in a multiple-document interface (MDI). /p /footnote - and runs the specified varcommand/var. + and runs the specified varcommand/var, + interpreting the entirity of the rest of the command + line as a command to pass straight to exec, in the + manner that ttxterm/tt does. /p/item itemp Support the command-line option tt-T
Bug#167604: debian-policy: provides the exception of static libraries.
Subject: debian-policy: provides the exception of static libraries. In Libraries section, In general, libraries must have a shared version in the library package and a static version in the development package. But, if libraries uses dlopen(3) internally, then some programs links the static version, it doesn't work correctly. because dlopen(3) doesn't try to resolve the symbols in the static-linked program. in most cases, the static version of such libraries is meaningless to be provided. You need to link your executable binary with -export-dynamic: man ld: -export-dynamic When creating an ELF file, add all symbols to the dynamic symbol table. Normally, the dynamic symbol table contains only symbols which are used by a dy namic object. This option is needed for some uses of dlopen. So I think our policy needs to be added the exception of such case. I do not agree, for the reason mentionned above. Cheers, Bill. pgpAbSXvhIIvt.pgp Description: PGP signature
Re: Looking at the 3 year olds
On Thu, Nov 14, 2002 at 06:47:35PM -0600, Manoj Srivastava wrote: * #39830: [AMENDMENT 30/10/2002] get rid of undocumented(7) symlinks Package: debian-policy; Severity: wishlist; Reported by: [EMAIL PROTECTED]; 3 years and 151 days old. Heh. Have people tested the new mandb? This is going to get included into policy on next upload; the new and improved mandb ought to be in unstable before then. It's in unstable for all architectures except hurd-i386 and m68k now. Cheers, -- Colin Watson [EMAIL PROTECTED]
Bug#167604: debian-policy: provides the exception of static libraries.
On Fri, 15 Nov 2002 13:27:49 +0100,
BA == Bill Allombert [EMAIL PROTECTED] wrote:
Subject: debian-policy: provides the exception of static libraries.
In Libraries section,
In general, libraries must have a shared version in the
library package and a static version in the development
package.
But, if libraries uses dlopen(3) internally, then some
programs links the static version, it doesn't work
correctly. because dlopen(3) doesn't try to resolve the
symbols in the static-linked program. in most cases, the
static version of such libraries is meaningless to be
provided.
BA You need to link your executable binary with -export-dynamic:
BA man ld:
BA-export-dynamic
BA When creating an ELF file, add all symbols to the
BA dynamic symbol table. Normally, the dynamic symbol
BA table contains only symbols which are used by a dy
BA namic object. This option is needed for some uses
BA of dlopen.
No, it doesn't help.
try this case:
---Makefile
all: libfoo.so libfoo.a module.so t-shared t-static
test: all
LD_LIBRARY_PATH=. ./t-shared
LD_LIBRARY_PATH=. ./t-static
libfoo.so: libfoo.c
gcc -fPIC -c -o [EMAIL PROTECTED] $
gcc -shared -Wl,-soname,$@ -o $@ $
libfoo.a: libfoo.c
gcc -c -o [EMAIL PROTECTED] $
ar ru $@ [EMAIL PROTECTED]
module.so: module.c libfoo.so
gcc -shared -fPIC -Wl,-soname,$@ -o $@ -L. -lfoo $
module_s.so: module.c libfoo.a
gcc -shared -fPIC -Wl,-soname,$@ -o $@ $ libfoo.a
t-shared: t.c libfoo.so
gcc -rdynamic -ldl -L. -lfoo -o $@ $
t-static: t.c libfoo.a
gcc -static -export-dynamic -o $@ $ -L. -lfoo -ldl
clean:
rm -f libfoo.so libfoo.a *.o t-shared t-static module.so
---libfoo.c
#include string.h
#include libfoo.h
static char *foo_msg = NULL;
static char *bar_msg = NULL;
void foo_init (char *s)
{
foo_msg = (char *)strdup (s);
}
char *foo_get (void)
{
return foo_msg;
}
void bar_init (char *s)
{
size_t len;
if (!foo_msg) {
printf (foo_msg isn't initialized.\n);
return;
}
len = strlen (foo_msg) + strlen (s) + 3;
bar_msg = (char *)malloc (sizeof (char) * len);
sprintf (bar_msg, %s: %s, foo_msg, bar_msg);
}
char *bar_get (void)
{
return bar_msg;
}
---libfoo.h
void foo_init (char *s);
char *foo_get (void);
void bar_init (char *s);
char *bar_get (void);
---module.c
#include libfoo.h
void plugin_init (char *s);
char *plugin_get (void);
void plugin_init (char *s)
{
bar_init (s);
}
char *plugin_get (void)
{
return bar_get ();
}
---t.c
#include stdio.h
#include dlfcn.h
#include libfoo.h
int main (int argc, char **argv)
{
void *handle;
void (*init)(char *);
char *(*get)(void);
char *error;
char *msg = test.;
char module[256];
foo_init (msg);
printf (%s\n, foo_get ());
if (argc 1) {
if (strlen (argv[1]) = 256) {
printf (too long\n);
exit (1);
}
strcpy (module, argv[1]);
} else {
strcpy (module, module.so);
}
handle = dlopen (module, RTLD_LAZY);
if (handle) {
init = dlsym (handle, plugin_init);
if ((error = dlerror ()) != NULL) {
printf (error: %s\n, error);
}
init (msg);
get = dlsym (handle, plugin_get);
if ((error = dlerror ()) != NULL) {
printf (error: %s\n, error);
}
printf (%s\n, get ());
dlclose (handle);
}
return 0;
}
then, runs make and make test.
This is the simple codes for reproducing a problem which
GTK+ has. but I don't think this is GTK+ specific problem,
because as I said, it's actually dlopen(3) issue.
--
Akira TAGOH : [EMAIL PROTECTED] / Japan GNOME Users Group
[EMAIL PROTECTED] : [EMAIL PROTECTED] / GNOME-DB Project
: [EMAIL PROTECTED] / Red Hat, Inc.
: [EMAIL PROTECTED] / Debian Project
Re: Bug#167422: files in /usr/share should be world-readable
On Fri, Nov 15, 2002 at 02:13:23AM -0600, Manoj Srivastava wrote: If these were ordinary log files of a running program generating data, I may agree. But since we are down to gut feelings, a bunch of files are generated by postinst, and they all reside in the same place. Some of these files are read by emacs; some times there is a README read by the inquisitive users, and then there is a record of what transpired while gemerating these files. It strikes my gut as sane that these files, all generated by the postinst, live in the same location. [...] They are not even accessed by the user directly, and certainly not by emacs. This description unambiguously screams /var/log to me. Clearly your mileage varies. -- G. Branden Robinson| Debian GNU/Linux | Please do not look directly into [EMAIL PROTECTED] | laser with remaining eye. http://people.debian.org/~branden/ | pgpAFxq6xFn9a.pgp Description: PGP signature
Bug#165063: debian-policy: Section `12.8.3 Packages providing a terminal emulator' fails to sufficently document the -e option
On Fri, Nov 15, 2002 at 09:41:15AM +, Jonathan David Amery wrote: I suggest that the following patch be applied to policy in order to resolve this issue: --- policy.sgml.old Fri Nov 15 09:30:47 2002 +++ policy.sgml Fri Nov 15 09:36:16 2002 @@ -6982,9 +6982,12 @@ emulator application were so coded, be a new view in a multiple-document interface (MDI). /p /footnote - and runs the specified varcommand/var. + and runs the specified varcommand/var, + interpreting the entirity of the rest of the command + line as a command to pass straight to exec, in the + manner that ttxterm/tt does. /p/item itemp Support the command-line option tt-T Seconded. -- G. Branden Robinson|I'm sorry if the following sounds Debian GNU/Linux |combative and excessively personal, [EMAIL PROTECTED] |but that's my general style. http://people.debian.org/~branden/ |-- Ian Jackson pgpYD5GJ3sYeB.pgp Description: PGP signature
