contemplations of libelf
Well, after a lot of fiddling and hacking and threatening of dpkg, I finally managed to get libelf compiling with 2.1.1.0 compliant sources. Before I upload it, though, I want a few things cleared up: 1) Should I rename the package to "libelf0" (Replaces: and Conflicts: libelf) in the same way that libc has libc4 and libc5? 2) Since libelf has a couple of header files also found in libc (which really shouldn't be there IMO, since libc doesn't implement those functions), I've added a Replaces: libc. Is this a Bad Thing(tm)? (it seemed the right thing to do when I was poking around the policy/ programmers' manual...) I'd appreciate responses; I don't want to tread on anybody's toes here...
Bug#4501: util-linux does not install the whereis command
On Sun, 15 Sep 1996, Christoph Lameter wrote: > util-linux does not install the whereis command from the utils. The > whereis command is used to find the location of a Unix command and is very > important. Actually the whereis command is pretty useless because the places it looks are hardcoded in. Much better is locate. I don't mind including it if you insist. > Reviewing the source package I noted that a large number of other tools > are also not installed. What are the reason for this? They are installed from other sources. util-linux is a really ugly package. Whenever possible, maintainers tried to find the upstream packages for the tools and use them. That way new versions and upstream bugs get fixed. Guy
Bug#4603: linux.cf should build libraries as Elf by default
Package: xdevel Version: 3.1.2-3 In order to have shared libraries built as ELF by default, the file /usr/X11R6/lib/X11/config/linux.cf should have the symbols LinuxElfDefault and UseElfFormat defined to YES by default. Yves. -- Yves Arrouye Email: [EMAIL PROTECTED] 7, avenue Leon BolleeWeb: http://www.fdn.fr/~yarrouye/ 75013 Paris Work: +33 45 95 64 59 France Home: +33 53 61 09 55
Bug#4604: Small typo in postinst of dwww_1.0-1
Package: dwww Version: 1.0-1 There is a small typo in the postinst script of dwww-1.0. Here is the appropriate fix: --- dwww.postinst.orig Fri Sep 27 11:57:56 1996 +++ dwww.postinst Fri Sep 27 11:58:05 1996 @@ -18,7 +18,7 @@ then echo "Linking dwww to Apache httpd." dolinks /var/web/webspace /var/web/cgi-bin - foundhttp=yes + foundhttpd=yes fi if test -x /usr/sbin/cern-httpd Dominik
Inappropriately closed bug
reopen 1708 quit I'm reopening Bug#1708 "`passwd' not interruptible when invoked by `adduser'". Christoph Lameter closed it with the comment "bug has nothing to do with adduser but is a passwd problem", which is not an appropriate reason for closing a bug report. Ian. (This message is BCC'd to debian-bugs-control.)
`eval' patch to dpkg-buildpackage
Please REVERSE the change you made to make the default rootcommand be `eval'. Please DO NOT ATTEMPT to make `su -c' work as a gain-root-command. It is IMPOSSIBLE to produce a correct implementation which works with both `su -c' and also with `really', `sudo', `super', &c. This is because `su -c' expects the command and all its arguments as a shell expression whereas all the rest expect the command and its arguments as separate arguments to themselves. Please see the manpage for dpkg-buildpackage and make sure that it implements what the manpage in 1.4.0 says: -rgain-root-command When dpkg-buildpackage needs to execute part of the build process as root, it prefixes the command it executes with gain-root-command if one has been specified. gain-root-command should be the name of a program on the PATH and will get as arguments the name of the real command to run and the arguments it should take. gain-root-command should not con- tain spaces or any other shell metacharacters. gain-root-command might typically be sudo, super or really. su is not suitable, since it requires a -c option to run a command and even then it can only invoke the user's shell with -c instead of passing arguments individually to the command to be run. There are several solutions that wouldn't necessarily break anything: 1. Write a wrapper script for `su' which does shell-quoting &c, so that it can be used like really, sudo and super but uses su -c internally. Getting this right so that it works even if the arguments have funny characters in is hard, but it happens that dpkg-source doesn't usually pass such arguments so you're probably OK with this application. However, don't distribute the script unless you've got it right. dpkg-buildpackage might at some point in the future build packages with arguments as specified. 2. Make su have an option to run a command specified with its arguments in separate arguments, using exec directly rather than " -c". You'd have to make sure that the user's shell was in /etc/shells for this to be safe. 3. Switch from using su to using something else. 4. Make dpkg-buildpackage have a different style of -r option which does what you want. This will quickly turn dpkg-buildpackage into a mess, so please don't do it. I do not have time to have a flamewar about this. Bruce and everyone else, please can you ensure that my instructions are carried out. If someone thinks that I'm making a mistake with this I want them to convince me of it BEFORE acting, because it seems that there is so much misunderstanding here. Thanks, Ian.
More dpkg stuff
Oh dear. Heiko Schlitterman says that he's changed some programs (he doesn't say exactly which) to change the ownership of debian/files and debian/substvars. This is not a good idea, and doesn't correctly solve the problem either. I think that the right fix is to put near the top of dpkg-buildpackage case `umask` in *77)echo >&2 \ 'dpkg-buildpackage: warning: bad umask will probably break build' ;; esac Ian.
More dpkg stuff
I see from Heiko Schlitterman's patch that he's added an option to change the architecture in the name of the .changes file made by dpkg-buildpackage. Why was this done ? I'm also not convinced that his implementation of the way -m and so forth are handled by dpkg-buildpackage will work if the values contain spaces. Please test this, and if it doesn't work go back to the way I intended doing it, with `set' and `"$@"'. I see also from his patch that he's removed the commentary about not using `su' in dpkg-buildpackage, but failed to change the formal specification of what happens. That this has happened is because the new way doesn't have a sensible formal specification, of course. As I said, please put it back the way it was and fix su instead of dpkg-buildpackage. His patch to convert backslashes will allow me to make a Debian source package which will execute a command of my choice when unpacked. This needs to be fixed. Use of Perl's `eval' on the filename which came out of tar is not safe. Regarding the other changes: the fixes to hardlink handling, Karl Sackett's error message bugfix, dpkg-name being moved and the changes from mv to mv -f in some places are fine. Thank you. Sorry if I seem ungrateful - I'm not. It's just that I have some definite ideas about the way certain things ought to be done - argument unparsing and the handling of arguments with spaces and so forth is one of them, and security aspects are another. I'd like to see the dpkg tools continue to be consistent with my goals of being secure and sane in the face of arguments which contain any characters. Ian.
Bug#4501: util-linux does not install the whereis command
On Fri, 27 Sep 1996, Guy Maor wrote:
maor>On Sun, 15 Sep 1996, Christoph Lameter wrote:
maor>
maor>> util-linux does not install the whereis command from the utils. The
maor>> whereis command is used to find the location of a Unix command and is
very
maor>> important.
maor>
maor>Actually the whereis command is pretty useless because the places it
maor>looks are hardcoded in. Much better is locate. I don't mind including
maor>it if you insist.
locate shows huge amounts of information and does not restrict the search
to sensible locations.
I can do
whereis ls
to find ls + manpages in standard locations.
if I do
locate ls
then the screen will be scrolling with lots of other stuff.
Also whereis is commonly available in other linux distributions.
maor>
maor>> Reviewing the source package I noted that a large number of other tools
maor>> are also not installed. What are the reason for this?
maor>
maor>They are installed from other sources. util-linux is a really ugly
maor>package. Whenever possible, maintainers tried to find the upstream
maor>packages for the tools and use them. That way new versions and
maor>upstream bugs get fixed.
But I noticed that a lot of the tools are NOT in other debian packages.
{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}
{}Snail Mail: FTS Box 466, 135 N.Oakland Ave, Pasadena, CA 91182{}
{}FISH Internet System Administrator at Fuller Theological Seminary {}
{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}
PGP Public Key = FB 9B 31 21 04 1E 3A 33 C7 62 2F C0 CD 81 CA B5
Bug#4606: rxvt binary should be suid, in /usr/X11R6/bin
Package: rxvt Version: 2.18-1 1. rxvt binary should be suid. Otherwise it can't make changes to utmp. 2. rxvt binary should be in /usr/X11R6/bin, not /usr/bin Guy
Bug#4501: util-linux does not install the whereis command
On Fri, 27 Sep 1996, Christoph Lameter wrote: > [justification for including whereis] ok, I'll include it in the next version. > But I noticed that a lot of the tools are NOT in other debian packages. Which ones, specifically? Guy
libident_0.17-4 uploaded to master
-BEGIN PGP SIGNED MESSAGE- Format: 1.5 Date: Fri, 27 Sep 1996 09:26:20 -0400 Source: libident Binary: libident Architecture: source i386 Version: 0.17-4 Distribution: unstable Urgency: low Maintainer: Dale Scheetz <[EMAIL PROTECTED]> Description: libident - a simple RFC1413 client library Changes: libident (0.17-4) unstable; urgency=low . * changed architecture from all to i386 Files: bda2b45b7a695a0c6a88fd936a581e85 557 devel extra libident_0.17-4.dsc 0173027eb1c6c7c481e5c38a82edc898 2448 devel extra libident_0.17-4.diff.gz 8761293e90935613f3c503b1baeeb869 11686 devel extra libident_0.17-4_i386.deb -BEGIN PGP SIGNATURE- Version: 2.6.2 iQB1AwUBMkvbk5Eoh1qC3Q5lAQE7TwL/cMON+7v9V4KwJGp7I7YVMn+UOOiUsRfY 0KEDxyn3dF0qI4NoWMf5rM+XFnW2LC2Zw0IlIl6ws2Lk2wlhgZzOxbTuD3PGlPkb MN5lfK7dYRdWO6/nVmybW7njkWV/+Jl1 =DDlf -END PGP SIGNATURE- Luck, Dwarf -- aka Dale Scheetz Phone: 1 (904) 877-0257 Flexible Software Fax: NONE Black Creek Critters e-mail: [EMAIL PROTECTED] If you don't see what you want, just ask --
Bug#4607: passwd segfaults if no finger info specified
Package: passwd Revision: 1.0-5 On one of my systems, passwd will segmentation fault after entering the new password if no finger information is present for the user in /etc/passwd. Typescript: Script started on Sat Sep 28 00:28:35 1996 > passwd Current password: New password (? for help): Segmentation fault > ^Dexit Script done on Sat Sep 28 00:28:46 1996 A user entry from /etc/passwd; hamish:sX7ugAy3wuats:1001:1001::/home/hamish:/usr/bin/tcsh when hamish changes his password, passwd will segfault as above. No problem if root changes hamish's password. I am unable to duplicate this on my own Debian system. passwd is 1.0-5, libc is 5.2.18-9. Hamish Moffatt, [EMAIL PROTECTED]
Bug#4608: How to use it?
package: mbr version: 1.0.0-2 I looked through the package and could not find anything on how to use it or install it. There is a mention of "make install" but since the sources are not included in the binary no installation is possible. Please include documentation on how to use it.
Re: `eval' patch to dpkg-buildpackage
Ian Jackson wrote:
: Please REVERSE the change you made to make the default rootcommand be
: `eval'.
: Please DO NOT ATTEMPT to make `su -c' work as a gain-root-command. It
: is IMPOSSIBLE to produce a correct implementation which works with
: both `su -c' and also with `really', `sudo', `super', &c.
... and I (somebody else) was happy, to have
dpkg-buildpackge run with and without the -r'su root -c' option ...
Zu frueh gefreut ...
: This is because `su -c' expects the command and all its arguments as a
: shell expression whereas all the rest expect the command and its
: arguments as separate arguments to themselves.
Hmmm.
: implements what the manpage in 1.4.0 says:
:-rgain-root-command
: When dpkg-buildpackage needs to execute part of the
...
: There are several solutions that wouldn't necessarily break anything:
:
: 3. Switch from using su to using something else.
Just trying ... sudo can't cope with relative path names (as needed to
run ./debian/rules), with dpkg-buildpackage -rsuper the dpkg-shlibdeps
fails (insecure $ENV{PATH} ...)
--> it seems all *.pl scripts detect the run as suid and refuse
operation, unless they get something like
$ENV{PATH} = "/usr/bin:/bin:/usr/sbin:/sbin";
$ENV{IFS} = " \t\n";
included.
: 4. Make dpkg-buildpackage have a different style of -r option which
: does what you want. This will quickly turn dpkg-buildpackage into a
: mess, so please don't do it.
Uh no, I think, that not what is intended ...
Heiko
--
email : [EMAIL PROTECTED] [EMAIL PROTECTED] [EMAIL PROTECTED]
pgp : A1 7D F6 7B 69 73 48 35 E1 DE 21 A7 A8 9A 77 92
finger: [EMAIL PROTECTED] [EMAIL PROTECTED]
Re: More dpkg stuff
Ian Jackson wrote: : : Oh dear. : : Heiko Schlitterman says that he's changed some programs (he doesn't : say exactly which) to change the ownership of debian/files and : debian/substvars. : case `umask` in : *77)echo >&2 \ : 'dpkg-buildpackage: warning: bad umask will probably break build' : ;; : esac Why? If we can handle such case (umask != *77), why we don't handle this? Grep in scripts/ for 'fowner' controllib.pl:@fowner = (getpwnam(getlogin()))[2,3]; dpkg-distaddfile.pl:chown(@fowner, "$fileslistfile.new") dpkg-gencontrol.pl:chown(@fowner, "$fileslistfile.new") dpkg-shlibdeps.pl:chown(@fowner, "$varlistfile.new") || I think, this behaviour is that, what would be expected. Please correct me, If I'm wrong. Heiko -- email : [EMAIL PROTECTED] [EMAIL PROTECTED] [EMAIL PROTECTED] pgp : A1 7D F6 7B 69 73 48 35 E1 DE 21 A7 A8 9A 77 92 finger: [EMAIL PROTECTED] [EMAIL PROTECTED]
Bug#4607: passwd segfaults if no finger info specified
On Sat, 28 Sep 1996, Hamish Moffatt wrote: > On one of my systems, passwd will segmentation fault after entering > the new password if no finger information is present for the user > in /etc/passwd. Typescript: > > [...] > > I am unable to duplicate this on my own Debian system. I can't duplicate it either. Could you mail me the uuencoded gzipped core dump in private email? Guy
Kernel packaging issues
I noticed that we do not use the kernel packages of the debian project in
our school. Every machine still compiles ones own kernel.
I thought about the reasons for it. Having the packages would be much
easier to handle I guess:
1. Kernel packages are not up to date and kept bug free. There is a
longstanding bug with NFS not working under the precompiled 2.0.6 kernels
that derailed us. Ever since the assumption is "Debian Kernels do not
work". NFS is essential for a campus net such as ours.
2. Kernel packages do not fit the machine. Notably the Network Servers
need a lot of functionality in the network area: Firewalling, bridging, IP
masquerading and the whole lot. Workstations/LAPTOPs in turn have a lot of
SCSI drivers included which are giving strange messages on bootup.
3. The use of kernel dependent modules (iBCS,ELKS,PCMCIA,electric fence
etc) is a pain since we have to recompile those souce modules separately
for each version.
Could we do the following:
1. Have one source code module for the kernel. Add to that all the
sourcecode for modules (iBCS,ELKS,PCMCIA,electic fence etc).
That way all the kernel version dependant stuff can be compiled together
and there is less of a chance of something being out of sync.
updating the kernel binary will lead to the upgrade of all the kernel
dependant modules and we wont have those versions mismatches anymore.
2. Remove the version number from the kernel package (unnecessary since
all kernel dependant modules are included in standard cases). Instead have
different kernels according to their purpose of use generated from one
source module
kernel-workstation All mouse drivers, IDE only and low-level
Network cards compiled it. No advanced networking
functions.
kernel-server Compiled in standard SCSI+IDE. High-level network
drivers. All advanced networking functions
compiled in.
kernel-limited Special kernel for low memory
situations/installation. Contains
minimal amount of drivers compiled in to have a
working kernel. SCSI can be insmoded during
install if necessary. The Base package should
include the full kernel instead of the bootdisks.
And be sure that the kernel package is immediately fixed by having a
number of people able to update the kernel package.
After the 2.0.X is abandoned by the developers (may be at 2.0.22) we
should be open to fixing problem in the kernel sourcecode so that not each
of us ends up with a "stable" kernel and lots of local patches like it
happened with 1.2.13.
{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}
{}Snail Mail: FTS Box 466, 135 N.Oakland Ave, Pasadena, CA 91182{}
{}FISH Internet System Administrator at Fuller Theological Seminary {}
{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}
PGP Public Key = FB 9B 31 21 04 1E 3A 33 C7 62 2F C0 CD 81 CA B5
Bug#4609: csh bombs when invoked from subdir w/ 28-char name
Package: csh Version: 5.26-3 I found not one but two bugs in /bin/csh. The incredibly bizarre symptom (on my system) is that if you try to run csh while cd'ed to a subdirectory of your home directory whose full pathname is exactly 28 characters long, csh dies with a segmentation fault. Using Electric Fence, it was pretty quick to track this down to a problem with s_strncmp (strncmp for short strings) which causes it to return "no match" for the case where the first string is an initial substring of the second, plus a malloc in dcanon that allocates one Char fewer than it should. Here are the fixes: *** /tmp/str.c.orig Thu Sep 26 12:35:25 1996 --- /tmp/str.c Thu Sep 26 12:35:25 1996 *** *** 338,343 --- 338,345 * negative collate low against normal characters but high against the * end-of-string NUL. */ + if (n == 0) + return(0); if (*str1 == '\0' && *str2 == '\0') return (0); else if (*str1 == '\0') *** /tmp/dir.c.orig Thu Sep 26 12:35:59 1996 --- /tmp/dir.c Thu Sep 26 12:35:59 1996 *** *** 876,882 */ p2 = cp + Strlen(p2); sp = newcp = (Char *) xmalloc((size_t) ! ((cc + Strlen(p2)) * sizeof(Char))); while (*p1) *sp++ = *p1++; while (*p2) --- 876,882 */ p2 = cp + Strlen(p2); sp = newcp = (Char *) xmalloc((size_t) ! ((cc + Strlen(p2) + 1) * sizeof(Char))); while (*p1) *sp++ = *p1++; while (*p2) These patches are relative to the sources on the I-Connect Debian 1.1 CD. I'm now using a custom-built 2.0.21 kernel. Both bugs have already been fixed in both tcsh and the FreeBSD version of csh. Look carefully at the s_strncmp fix, I think FreeBSD csh and tcsh have a better solution than mine. -Randy -- http://cogsci.ucsd.edu/~gobbel/
gmp_1.3.2-5 uploaded to master
-BEGIN PGP SIGNED MESSAGE- Format: 1.5 Date: Fri, 27 Sep 1996 11:34:36 -0400 Source: gmp Binary: gmp Architecture: source i386 Version: 1.3.2-5 Distribution: unstable Urgency: low Maintainer: Dale Scheetz <[EMAIL PROTECTED]> Description: gmp- Multiprecision arithmetic library Changes: gmp (1.3.2-5) unstable; urgency=low . * fixed debian/rules to get so numbers from the changelog Files: 5b628a783da993f0dc505d91d0dca4c4 541 devel Standard gmp_1.3.2-5.dsc cbaeee45eb204edee0c2b117196b4845 3398 devel Standard gmp_1.3.2-5.diff.gz 9f171ac0909d47e87e90ceca3cc20840 46722 devel Standard gmp_1.3.2-5_i386.deb -BEGIN PGP SIGNATURE- Version: 2.6.2 iQB1AwUBMkwFlZEoh1qC3Q5lAQEHgQL/bmtY+xtBABEClSPl/BrrheSnyRIQxKmN WoPjr72SWXMIkpddKnCF/XtcwVsBc3xt5/NyjvLvx+4yyiFpgTIir6l1dvMegOvK GdcgyT9rnfRkNsJgAxw2J/xhazSxuc9W =k01D -END PGP SIGNATURE- Luck, Dwarf -- aka Dale Scheetz Phone: 1 (904) 877-0257 Flexible Software Fax: NONE Black Creek Critters e-mail: [EMAIL PROTECTED] If you don't see what you want, just ask --
Bug#4610: gcc tries to compile #ifndef'ed-out code
Package: gcc Version: 2.7.2.1-1 I was trying to track down a bug in csh (5.26-2), and so compiled with -DSYSMALLOC. Most of alloc.c is inside of an #ifndef SYSMALLOC conditional, but gcc complained about a bunch of stuff in alloc.c anyway: cd /usr/src/csh-5.26/ make gcc -O3 -I/usr/include/bsd -include /usr/include/bsd/bsd.h -DFILEC -DNLS -DSHORT_STRINGS -I. -DSYSMALLOC -c csh.c -o csh.o gcc -O3 -I/usr/include/bsd -include /usr/include/bsd/bsd.h -DFILEC -DNLS -DSHORT_STRINGS -I. -DSYSMALLOC -c alloc.c -o alloc.o alloc.c:447: conflicting types for `malloc' extern.h:318: previous declaration of `malloc' alloc.c:462: conflicting types for `realloc' extern.h:319: previous declaration of `realloc' alloc.c:477: conflicting types for `calloc' extern.h:320: previous declaration of `calloc' make: *** [alloc.o] Error 1 Compilation exited abnormally with code 2 at Fri Sep 27 10:33:06 -- The lines referred to in the error messages are well inside the #ifndef. Putting an #if 0 around the #ifndef'ed section made the complaints disappear. -Randy -- http://cogsci.ucsd.edu/~gobbel/
Re: `eval' patch to dpkg-buildpackage
On Fri, 27 Sep 1996, Heiko Schlittermann wrote:
> Ian Jackson wrote:
> : Please REVERSE the change you made to make the default rootcommand be
> : `eval'.
Yes, please do, dpkg-buildpackage 1.4.0.1 is *BROKEN* for me as I can
no longer use sudo to build my packages.
> Just trying ... sudo can't cope with relative path names (as needed to
> run ./debian/rules), with dpkg-buildpackage -rsuper the dpkg-shlibdeps
> fails (insecure $ENV{PATH} ...)
sudo 1.5-1 copes with relative path names. I've been able to build all
my packages with it.
Guy
Bug#4612: dhcpd overwrites existing config file /etc/dhcpd.conf
Package: dhcpd Version: 0.5.13-1 Dhcpd overwrites an existing /etc/dhcpd.conf file without warning. I suppose I have no one else but myself to blame for this one, it burnt me last upgrade as well, but I didn't report it as a bug. So it got me again this time. :-) Mark
Re: Bug#4612: dhcpd overwrites existing config file /etc/dhcpd.conf
You are right. Forgot to put the conffiles into the binary.
0.5.13-2 fixes this.
On Fri, 27 Sep 1996, Mark Purcell wrote:
msp>Package: dhcpd
msp>Version: 0.5.13-1
msp>
msp>Dhcpd overwrites an existing /etc/dhcpd.conf file without warning.
msp>
msp>I suppose I have no one else but myself to blame for this one, it burnt me
msp>last upgrade as well, but I didn't report it as a bug. So it got me again
msp>this time. :-)
msp>
msp>Mark
msp>
msp>
msp>
{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}
{}Snail Mail: FTS Box 466, 135 N.Oakland Ave, Pasadena, CA 91182{}
{}FISH Internet System Administrator at Fuller Theological Seminary {}
{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}
PGP Public Key = FB 9B 31 21 04 1E 3A 33 C7 62 2F C0 CD 81 CA B5
Bug#4501: util-linux does not install the whereis command
On Fri, 27 Sep 1996, Christoph Lameter wrote: > On Fri, 27 Sep 1996, Guy Maor wrote: > > maor>> But I noticed that a lot of the tools are NOT in other debian packages. > maor> > maor>Which ones, specifically? > Sorry I have to pass on that one right now. Leaving for a weekend retreat. > Next week if I have time I will hopefully be able to make a list. Actually I just made such a list by just doing a find on a compiled source tree, piping to type, and grepping for 'not found': ddate NO - silly dsplit split whereis YES? ctrlaltdel NO - obsolete sln NO - obsolete readprofile NO - obsolete setsid YES? cytune YES? rev tac dsplit and rev have the same functionality as split and tac provided by other packages. ddate is a bit too silly to have all users install it; ctrlaltdel, sln, and readprofile are obsolete; so whereis, setsid, and cytune are truly missing. I'll add whereis for the sake of completeness. cytune is a pretty specialized utility so I'll put in its own package, and setsid is of dubious use on a job control shell. I suppose I'll include it as it's tiny. Guy
Bug#4501: util-linux does not install the whereis command
On Fri, 27 Sep 1996, Guy Maor wrote:
maor>> But I noticed that a lot of the tools are NOT in other debian packages.
maor>
maor>Which ones, specifically?
Sorry I have to pass on that one right now. Leaving for a weekend retreat.
Next week if I have time I will hopefully be able to make a list.
{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}
{}Snail Mail: FTS Box 466, 135 N.Oakland Ave, Pasadena, CA 91182{}
{}FISH Internet System Administrator at Fuller Theological Seminary {}
{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}{}
PGP Public Key = FB 9B 31 21 04 1E 3A 33 C7 62 2F C0 CD 81 CA B5
Re: More dpkg stuff
On Fri, 27 Sep 1996, Heiko Schlittermann wrote: > Why? If we can handle such case (umask != *77), why we don't handle > this? > > Grep in scripts/ for 'fowner' > > controllib.pl:@fowner = (getpwnam(getlogin()))[2,3]; > dpkg-distaddfile.pl:chown(@fowner, "$fileslistfile.new") > dpkg-gencontrol.pl:chown(@fowner, "$fileslistfile.new") > dpkg-shlibdeps.pl:chown(@fowner, "$varlistfile.new") || > > I think, this behaviour is that, what would be expected. Please correct > me, If I'm wrong. This can break, for example: I login as user foo. I su to user bar. I build a package. Oops - now foo owns the files created by debian/rules. There's a danger in making things idiot-proof. In general, the simplest solution is the best one. Because the rules clean target should rm debian/substvars (if it's dynamic) and debian/files (prog manual 3.2.[45]), there's nothing wrong with having them owned by root. Ian's solution of warning about the umask is simplest. Add something like this to the top of debian/rules: expr $(umask -S): '.*o=r' > /dev/null && \ echo "Your umask will probably break dpkg-buildpackage" 2>&1 Guy
Bug#4613: Squid overwrites squid.conf
Package: squid Version: 1.0.16-1 Overwrites /etc/squid.conf, looks like squid.conffiles didn't make it into the binary. Mark
lesstif?
I have been looking for the source to the lesstif libraries so I can use them to build grok. So far all altavista searches have provided URLs that don't respond. Does anyone have any idea where this library can be found? (preferably source) Thanks, Dwarf -- aka Dale Scheetz Phone: 1 (904) 877-0257 Flexible Software Fax: NONE Black Creek Critters e-mail: [EMAIL PROTECTED] If you don't see what you want, just ask -- -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED]
Re: lesstif?
Don't you just hate it when people answer their own questions? I finally found what I needed (I hope) at ftp://ftp.uoknor.edu/linux/sunsite/X11/libs/ Sorry to be a bother, Dwarf -- aka Dale Scheetz Phone: 1 (904) 877-0257 Flexible Software Fax: NONE Black Creek Critters e-mail: [EMAIL PROTECTED] If you don't see what you want, just ask -- -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED]
Re: Bug#4501: util-linux does not install the whereis command
On Fri, 27 Sep 1996, Guy Maor wrote: > Actually I just made such a list by just doing a find on a compiled > source tree, piping to type, and grepping for 'not found': > > ddate NO - silly Yes, it's silly, but I used to have ddate show up now and then on login. Since debian didn't come with it, my rc script that used that got broken and had to be fixed. > other packages. ddate is a bit too silly to have all users install it; It's a tiny program. I think it's worth installing just for completeness and compatability with other distributions. -- #!/usr/bin/perl -i\$q='$q',\$p='$p';eval\$q.\$\^I\n"# # [EMAIL PROTECTED] $q='print"$p$^I\n',$p='#!/usr/bin/perl -i';eval$q.$^I # Joey Hess "He. He. He." - - Herman Toothrot -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED]
Re: lesstif?
I believe the canonical source for the lesstif libraries is ftp://ftp.hungry/pub/hungry/lesstif. There's a debianized version of lesstif in Incoming/ on master. It's based on the 0.50 release, though I doubt it'd be all that hard to upgrade to the in-progress version. -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED]
m4_1.4-5 uploaded to master
-BEGIN PGP SIGNED MESSAGE- Format: 1.5 Date: Fri, 27 Sep 1996 14:10:56 -0400 Source: m4 Binary: m4 Architecture: source i386 Version: 1.4-5 Distribution: unstable Urgency: low Maintainer: Dale Scheetz <[EMAIL PROTECTED]> Description: m4 - a macro processing language Changes: m4 (1.4-5) unstable; urgency=low . * Converted to Standards-Version 2.1.0.0 Files: 43338068258b424b3d837707bd2d6c34 531 devel standard m4_1.4-5.dsc 57db6042ce9cd8a1c997b558bd51618f 317927 devel standard m4_1.4.orig.tar.gz 608dd4d8e235435949bec4f35a58540f 2941 devel standard m4_1.4-5.diff.gz bd063ecfe74bb0f68082978bcc94df3b 76704 devel standard m4_1.4-5_i386.deb -BEGIN PGP SIGNATURE- Version: 2.6.2 iQB1AwUBMkwcwpEoh1qC3Q5lAQG/UAMAsF1vJaXZ0LORjTqTxFyNF+lAOcgHWcyw HcTH7TbrRyEjvHDixlhFUWJvzaMZ/gMLQDy4Vjnuxt3ygbxktFaGQp2v4Cm107s3 1zW2eVq8CcHyx2ePoUMSrdv4UtT0gDi9 =tBKu -END PGP SIGNATURE- Luck, Dwarf -- aka Dale Scheetz Phone: 1 (904) 877-0257 Flexible Software Fax: NONE Black Creek Critters e-mail: [EMAIL PROTECTED] If you don't see what you want, just ask -- -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED]
