/etc/host.conf /etc/protocols, and /etc/services
Peter, Could you have "netbase" include /etc/host.conf, /etc/protocols, and /etc/services? I'm placing them in the base package for now, but they really don't belong there. If they are in the netbase package, they will be extracted from there and installed on the base floppies by the basedisks.sh script. Thanks Bruce -- -- Attention Radio Amateurs: For information on "Linux for Hams", -- read the WWW page http://www.hams.com/LinuxForHams, -- or e-mail the word "help" to [EMAIL PROTECTED]
dpkg approaching 1.0 and non-beta status
There are a number of things I want to do to dpkg, but I think they all count as extra features, rather than bugfixes. I therefore propose to call dpkg non-beta very shortly, and issue version 1.0.0. NB that dpkg version numbers will then no longer have any apparent resemblance to Debian release numbers. The resemblance was misleading, anyway, but I haven't been able to do much about it yet, because I couldn't call dpkg 1.0 until it wasn't beta, and I couldn't reduce the version number below 0.93.x, which is the number I inherited. Ian.
Re: /etc/host.conf /etc/protocols, and /etc/services
Bruce Perens wrote: > Peter, > > Could you have "netbase" include /etc/host.conf, /etc/protocols, and > /etc/services? I'm placing them in the base package for now, but they > really don't belong there. If they are in the netbase package, they will > be extracted from there and installed on the base floppies by the > basedisks.sh script. Can we assume that everybody installs netbase? If not you'll have problems if /etc/services is not there. Every program that does a getservbyname, getservbyport or getservent (e.g. syslogd) won't work. The same problem might exist with /etc/protocols and getprotoent, getprotobyname and getprotobynumber. Because of these reasons Ian M. put them on the basedisk. Peter -- Peter TobiasEMail: Fachhochschule Ostfriesland [EMAIL PROTECTED] Fachbereich Elektrotechnik und Informatik [EMAIL PROTECTED] Constantiaplatz 4, 26723 Emden, Germany
Bug#1513: /bin/kill segfaults if invoked instead of killall
Package: bsdutils
Version: 1.3-1
chiark:~> /bin/kill -HUP syslogd
Segmentation fault (core dumped)
chiark:~>
Ian.
begin 664 core.gz
M'XL(``,>;3`"`^U;;VPC1Q4?)^DUN1/4:7LH<)3;([EMAIL PROTECTED]/^F\29L$U\MA.[
MMXGW;.>2-&EJQ[N)%Z_ML.OD$JD-53FIE5I$4"L$0D7ECP1("-$O2(#`]X$/
MH%15):A0A72*3GPH4%6A1>TAE2SOS:X37R\M/55'0=U?]-MY\^;->[,[N[.[
M+^M7.$*^_'Q;KD8(Z?RG68>"W'W5*D\SFN9ITLED$]#U]Y9<-_D,N:/%1;K`
M[C1Q$Q>TZ:VF:[EMAIL PROTECTED]&?73[UE%;M>.Z=)WT;OC%RU?V1E#X7OS"E5M'6T!R
MV638->O'L&RSJGW?.DXZH;&D:EJSFZ[MBV2FM49:6BSYG;![V]OWP8$#!PX<
M.'[EMAIL PROTECTED]($#!PX<./C_QN\_];<]]G;>T]W+1:LK&[JZ7*QQ/85>SC\4#M\-VZ$@
MVX:.=F2+"I=6EI5*S>"J2UP-JE,5=4W1#;[EMAIL PROTECTED],<%]$TCGDS
M.%TQ%'U-D?FC+!"^F/.%CC`?Y'K.*'I)T92-7B[D#7B'0C"6*VY"[EMAIL PROTECTED](
M?`&X#?P-\!+PE\"[EMAIL PROTECTED]&O`)X*/`1X`/`=?=UOXV\@;M
M'[/J[:>:[EMAIL PROTECTED]&L+MR^?(.2U$TR1,UK:"-EMSV&E%?,$'"'BL8[?J"-FW)6`3AEB^4Y9WS&&\=1+4IYQST($#!PX<.'[EMAIL PROTECTED]($#!PX<
M.'#P[I")[EMAIL PROTECTED]&>H]7695\SZR=OWUULM`,DU"V`KM7V0R3+3TL[
M(<<^87W-(=OYD2<>[/=X`?D!OM,Z";MG5L3*#[Z9MF?[KHV/_-#%Y'<;*3`UYF_7%.SU.P#]P7C
MK]LY%DSY8#[G8E/.Q:NIB\@";U3Y$#_`!P:[EMAIL PROTECTED]/7P?Q:X?HK`C[`\
M-*RSDI4_WH7U%-[C?&=UEKL__`7;\FU09_0-TEN'[EMAIL PROTECTED]
MT&_)/?)<6P[O!:]P%F70'X5ZD5C?".(8+G=:7AK?"C9B(;4WS/JLV[KGR/::
[EMAIL PROTECTED]"RP"'P,^"1PY%]F_2SP/%`&_A7X!I#LF74W\"3P'/`^
MH`*L`G\.?`[X$G`'>!789IKU)/`\L`C<[EMAIL PROTECTED]/!YX)^`NV;C&'D7U8J7
M?2[8EYB2B+%A:-5EF8BI\%>]4/6N*+KF52L%;556
M2"(N2E(DFQ"\145;(5)D/)X&0T-G?C7%,[EMAIL PROTECTED]@PHSU/[EMAIL
PROTECTED]:T6ULFR02#J:
M$#2ULKI.LO'T1#0B"6O&@^LU12];V[Y`R!*,!]<,O\]G53CC]\#UOH-64=$L)[SR"*-V=($C44M2:*QE"6=H5.2)[EMAIL PROTECTED])RTI
M"MJ\AK)(\V6H+&K"PCB6!BW(J+^7%A0LS]("LTL``Z`L,P>J1Q[VR`E:J':'
M?="M8#2I=5JH836X#"UR`46)RLS+!&B*5>8/CQ-S%BI2166"1LLJJ$O,[EMAIL PROTECTED]
MCH)`2XI5T:S=+_FAFI)H*8#E.5H*8IFFI1"6&;3HQP[^_DU:&F#2($B#3`J#
M%&;2$$A#*`5\F]AGF,E^E,?\3`YNTK$`DT(@!9D$/L="3!H`B<4)A%D?%BD`
M7L=8I"!X&F.1@@&06*0@>HPP*<1B5KH#>.A&?6Q$,"`V-V!?BC$K-/H4-8F*1Y15A(@;:,_F":,!Q&
M-W!V!ZF.19@:2\)\A1I5VZ5AQS`P!LY\3;7/])'08)'66'/@7JN.9SB8K*Z@
M,D)7V3D0*M-5V\EZA1XEB60FFTK/"H'^`7;="NRJ)(E4)IN=E6#E"(8'B+4T
M6)<[6Q#P,J=-"[EMAIL PROTECTED]@[(E0ZL6\AJJJ;6QEY%]P3OC]S=;&JS1V#J1_9L\])C*P-J',<1X
M)B/[EMAIL PROTECTED]/RH/_$TG)V.I:<%3Y/HXC[S)R_.>
M&5Z\G^<[EMAIL PROTECTED]<%!Y$9<(HU+S<=U97F%Q)(928S,"M3'[EMAIL PROTECTED];GF2[
M:IOQ-;VB5M0:FR%[1YL&E8RF)H6F*!2&%GQ/8R/@#R=<9$NXM7/)F!`,]O>'
M0J%^$K_N0$_'&K,[EMAIL PROTECTED]>=%(4A2
MZ?'(9/*^2#8)0V71N*F*II;5FB*3>"P)[EMAIL PROTECTED]/,%N.6!YI?!__U/=/D&=^0L>3';.7F/#76[+?9#UM^-F
MQ^NZP?;W.[ZN__+^'%9W'=:^LVT]']IZ?$X]+EA$_.KS!WD+]HX.;`=V`.\$
M?AS8#>P!!H!H_A=XH/PME+_#]_=6*R_QOA^J+WW6XG7YE&O__A/:FN07'WL(
MQ_:E&^%^HL:FR_[52EZOD3XYKY3W+]*O''(A;I.NPP;E>YKXECK9L6]A?5^`
[EMAIL PROTECTED]@AO.]:O7'Z[_+YFF;,RQQ5+1_T?/!0K*36!]EQ]LT;^IO
MC="NT?=A<^1FO=/AO.S\>H_D_L"[WHN]^]M[^]P_A;?VG)=W!PX^3.!PX3Y@
LUOZM)N99[K+YEFG6;[=-.YNZ-A::B'[EMAIL PROTECTED]>Z7?B;R'\#V&[EMAIL PROTECTED]@
`
end
Bug#1514: bad texinfo entry in /usr/info/dir
Package: texinfo Version: 3.6 Revision: 3 /usr/info/dir contains the line (both with 3.6-2 and 3.6-3) * Texi: (texi).The GNU Project's documentation format. whereas the info files are called /usr/info/texinfo*. Substituting [Tt]exi to [Tt]exinfo by hand in the /usr/info/dir file permits to call the doc under its proper name, ie 'info texinfo'. -- [EMAIL PROTECTED] http://qed.econ.queensu.ca/~edd
jgraph
The jgraph program built with current tools segfaults before reaching main(). This applies to the current jgraph package. Rebuilding sources from an earlier package revision with a known working jgraph program in the binary package produces a broken jgraph program. Please withdraw the jgraph package from the distribution for now. [EMAIL PROTECTED] (Bill Mitchell)
Re: /etc/host.conf /etc/protocols, and /etc/services
Peter, Even if someone doesn't install netbase, the files in /etc from the netbase package will be part of the base floppy set (or the equivalent base tar archive used by the CD install) because of the script I use to create the base floppy set. So what I'm really asking for is for you to maintain those three files as part of your package, and I'll copy them from your package into the base system. Thanks Bruce
Bug#1497: acct-alpha-5-7 problems
This is a long, and somewhat belated, response. Please read through it and give me your comments. Marek Michalkiewicz writes: Marek> I noticed the following problems with the acct-alpha-5-7 package: Marek> Marek> - lastcomm and the kernel seem to disagree about sizeof(struct acct) Marek> (only every 13th line displayed looks like a valid one). After the Marek> (somewhat garbled) data from the accounting file, lastcomm displays Marek> several hundred lines of garbage, and finally gets a SEGV. I cannot reproduce the SEGV behaviour. But I remember that my machine behaved similarly at some point in time. This might be due to the fact that there is some old information in your /var/account/pacct file. Please set this file to zero, start accounting, execute some commands and try again. Output from `last' is not well formatted, a better `last' exists. See below. Marek> - /etc/init.d/acct prints the usage message when invoked correctly Marek> (with "start" or "stop" as the first argument), and doesn't start or Marek> stop accounting (but running accton by hand works as expected). It Marek> works after commenting out the ". /etc/init.d/functions" line (which Marek> is not necessary here anyway), so I guess maybe something is screwed Marek> up in this file (part of sysvinit). Thanks, I will suppress sourcing of /etc/init.f/functions in the next release. Marek> - trying to start accounting again when it is already running gives Marek> a message "Process accounting not available on this system", which Marek> is a bit confusing (I suggest not to redirect stderr from accton to Marek> /dev/null, so that one can see what is really the problem). Please see acct(2). There is not a lot we can do about it as the acct() functions returns either 0 or -1. You get -1 for missing accounting support in the kernel, but also when accounting is already turned on. All I could do is to change the wording of the message to Process accounting not available or already running. but this could irritate new users considerably. The REAL problem with acct-alpha is that it doesn't really match the current kernel. acct-alpha stems from a package that was released in 1993 with url ftp://alpha.gnu.ai.mit.edu:/gnu/acct-alpha-5.tar.gz. It provides ac report on connect time from `wtmp' accton turn on process accounting last list logins on system lastcomm prints information about previously executed commands sa summarizes info about previously executed commands but 'sa' can't compile under the current kernel (some struct members missing in a kernel header file). 'ac' feels buggy (creates output for dates not included in the wtmp file) and `lastcomm' output is not properly formatted. ftp://sunsite.unc.edu/pub/Linux/system/Admin/accounts/acct-1.2.0.tar.gz provides accton turn on process accounting accttrim trim down size of an account file lastcomm show last command executed in reverse order The programs work fine, and they have man pages. I could package acct-1.2.0 instead of acct-alpha and continue to package `last' as a single package for the base section. I already wrote a man page for `last'. We would lose ac, which doesn't work well right now, and sa, which doesn't work at all. But we would gain a proper, current `acct' package. Please comment. -- [EMAIL PROTECTED] http://qed.econ.queensu.ca/~edd
Bug#1515: mount -- wrong error msg
Package: mount version: 2.4 revision: 5 mount report a wrong error msg, if a user tries to re-mount a already mounted device: # uname -a Linux luna 1.2.13 #8 Mon Sep 4 10:50:08 MET DST 1995 i486 # mount --version (u)mount: version from util-linux-2.2 # grep /CD /etc/fstab /dev/scd0 /CD iso9660 ro,user,noauto # mount -v /CD /dev/scd0 on /CD type iso9660 (ro,noexec,nosuid,nodev) # mount /CD mount: /dev/scd0 already mounted or /CD busy This error msg is OK. # su - ke luna:/home/karl$ mount -v /CD mount: only root can mount /dev/scd0 on /CD *** Not OK. luna:/home/karl$ exit logout # umount /CD # su - ke luna:/home/karl$ mount -v /CD /dev/scd0 on /CD type iso9660 (ro,noexec,nosuid,nodev) luna:/home/karl$ mount -v /CD mount: only root can mount /dev/scd0 on /CD *** Same bug.
Bug#1516: 27 Sep (I think) root disk bug
When an error is encountered reading a base disk (the error I got was "Unexpected EOF"), the user is dumped quickly back to the menu. The install program should pause with the error message displayed on the screen, and give the user a chance to read it and press to dump back to the menu. [EMAIL PROTECTED] (Bill Mitchell)
Re: /etc/host.conf /etc/protocols, and /etc/services
Bruce Perens wrote: > Peter, > > Even if someone doesn't install netbase, the files in /etc from the netbase > package will be part of the base floppy set (or the equivalent base tar > archive used by the CD install) because of the script I use to create the > base floppy set. So what I'm really asking for is for you to maintain those > three files as part of your package, and I'll copy them from your package > into the base system. Sure, no problem. They will be part of the next netbase package (as conffiles). BTW: Can you send me your current versions of these files (especially /etc/services and /etc/protocols)? Peter -- Peter TobiasEMail: Fachhochschule Ostfriesland [EMAIL PROTECTED] Fachbereich Elektrotechnik und Informatik [EMAIL PROTECTED] Constantiaplatz 4, 26723 Emden, Germany
0.93r6 install procedure problem
After the initial install, there's a prompt to set up the root password, then a drop into dselect. If, while the dselect session is going on, a login as root is done on another VC, that login also gets a prompt to set up root and a drop into dselect. The initial root login sequence is changed when either of the two dselect dropins exits. The root login sequence should be changed before the dropin to dselect. [EMAIL PROTECTED] (Bill Mitchell)
0.93r6 install sequence suggestions
First, terminology. There's Normal vs. Expert mode. I suggest that this be changed to Newuser vs. Expert. I'll use that terminology below. That's suggestion #1. #2 follows. After install, there's a dropin to dselect. dselect presents an interface similar in complexity to the initial install Expert mdoe. It seems that if there's a Newuser mode for install, a similar Newuser mode is needed for at least this initial dselect dropin. Without looking at either the first-login stuff in detail, or looking at the dselect code, it seems that this might not be very difficult to achieve. 1. The dselect dropin might prompt for Newuser or Expert mode (or might have this info passed in via an installed file). 2. Expert mode would be as presently done. 3. Newuser mode would need support from dselect so menu choices [A], [S], [I], and [Q] could be done in sequence without presenting the dselect menu to the user. Should be simple to do, except possibly for going direct to [Q] if [A] exited without establishing access. [EMAIL PROTECTED] (Bill Mitchell)
dselect dies in getgrnam()
I'm sorry that I wrote TarExtractor to naively call getgrnam() when the input group name was invalid. I'm patching my version to not do that - would you do so as well, Ian? Thanks Bruce
Re: dselect dies in getgrnam()
Bruce Perens writes ("dselect dies in getgrnam()"):
> I'm sorry that I wrote TarExtractor to naively call getgrnam() when
> the input group name was invalid. I'm patching my version to not do
> that - would you do so as well, Ian?
Yes, that would be good, thanks.. If you send me a patch I'll include
it. If you don't have time and the solution will be obvious (I
haven't looked at the TarExtractor code in this area) don't bother and
I'll do it myself.
Obviously the libc bug should be fixed too. There are two open bugs
for this problem, and I suggest we leave at least one them open until
the libc is OK.
Ian.
