Re: rc & runsystem

2004-12-25 Thread Alfred M. Szmidt
"Configuration file" is a Debian thing, not a GNU thing. You are applying the Debian package management standards to something which is actually not Debian. Dear me, why must you kick a dead horse? And kicking it in the wrong place at that... We do actually have "configuration files",

Re: rc & runsystem

2004-12-25 Thread Alfred M. Szmidt
Oh dear, oh dear... Who let Thomas escape from his little solitary prison? He could hurt himself you know. ___ Bug-hurd mailing list Bug-hurd@gnu.org http://lists.gnu.org/mailman/listinfo/bug-hurd

Re: rc & runsystem

2004-12-18 Thread Thomas Bushnell BSG
"Alfred M. Szmidt" <[EMAIL PROTECTED]> writes: > But there are perfectly valid reasons to edit /libexec/rc, and as such > it should be classifed as a configuration file. If so, it should get > the same treatment as /etc/motd and friends get right now. "Configuration file" is a Debian thing, not

Re: rc & runsystem

2004-12-18 Thread Thomas Bushnell BSG
"Alfred M. Szmidt" <[EMAIL PROTECTED]> writes: > No, I won't get used to it since the behaviour of `make install' is > wrong. That _I_ use it as a package manager is totaly irrelevant to > the dicussion. Clobbering files that *will and should* get modifed by > users is wrong. You are applying t

Re: rc & runsystem

2004-12-17 Thread Alfred M. Szmidt
Does make dist even work? Mmm... Nope. Sighs... I'm going to bed now... Attached the wrong log; here is the correct one. make -C libpthread lndist no_deps=t make[1]: Entering directory `/home/ams/hurd/libpthread' mkdir ../hurd-snap/libpthread ln ./Makefile ./ChangeLog ./pt-attr.c ./pt-attr-d

Re: rc & runsystem

2004-12-17 Thread Alfred M. Szmidt
On second thought, that breaks make dist, possibly among other things. Just commit your original config/Makefile patch. Done. Does make dist even work? Mmm... Nope. make -C libshouldbeinlibc lndist no_deps=t make[1]: Entering directory `/home/ams/hurd/+build/libshouldbeinlibc' ln ../../li

Re: rc & runsystem

2004-12-17 Thread Roland McGrath
>I guess the right thing would be to exclude it always and just have >a make install-config target that does make -C config install. > > Untested. > > 2004-12-18 Alfred M. Szmidt <[EMAIL PROTECTED]> > > * Makefile (other-subdirs): `config' removed. > (install-config): New t

Re: rc & runsystem

2004-12-17 Thread Roland McGrath
I would be perfectly fine with saying that runsystem, runttys, and rc, are outside the scope of the core hurd package and should be separate (along with motd and ttys). They can be in a separate source tarball, and that would be a package that noone would use if there is a better replacement to be

Re: rc & runsystem

2004-12-17 Thread Alfred M. Szmidt
The /etc/login files are in the grey area, like rc would be if it stilled lived in /etc. ;-) O... Lets move it back. And remove that little executable bit, and make runsystem do "/bin/sh /etc/rc". I don't know if I'm serious or not about that; since rc doesn't have to be written in sh...

Re: rc & runsystem

2004-12-17 Thread Alfred M. Szmidt
I guess the right thing would be to exclude it always and just have a make install-config target that does make -C config install. Untested. 2004-12-18 Alfred M. Szmidt <[EMAIL PROTECTED]> * Makefile (other-subdirs): `config' removed. (install-config): New target. Index:

Re: rc & runsystem

2004-12-17 Thread Roland McGrath
>Was there anything wrong with this patch? > > Meh, late... This patch being the following, I think we can all agree > that /etc only contains configuration files. The /etc/login files are in the grey area, like rc would be if it stilled lived in /etc. ;-) I think it would be better to decr

Re: rc & runsystem

2004-12-17 Thread Roland McGrath
> But who knows, I might do that if people who actually use the Hurd > (i.e. not you) pester about it long enough. :-) That is usually the best way to prioritize the work items. ___ Bug-hurd mailing list [EMAIL PROTECTED] http://lists.gnu.org/mailman/l

Re: rc & runsystem

2004-12-17 Thread Alfred M. Szmidt
>It's still not a configuration file. > > You are wrong. I'm glad we can agree on that, pronoun binding notwithstanding. I think we can move on now to where you send a patch to add a nondefault option for the behavior you want. I will leave that as an exercise for someone

Re: rc & runsystem

2004-12-17 Thread Roland McGrath
>It's still not a configuration file. > > You are wrong. I'm glad we can agree on that, pronoun binding notwithstanding. I think we can move on now to where you send a patch to add a nondefault option for the behavior you want. Thanks, Roland

Re: rc & runsystem

2004-12-17 Thread Roland McGrath
> >Was there anything wrong with this patch? > > > > Meh, late... This patch being the following, I think we can all agree > > that /etc only contains configuration files. > > The /etc/login files are in the grey area, like rc would be if it stilled > lived in /etc. ;-) I think it would be

Re: rc & runsystem

2004-12-17 Thread Alfred M. Szmidt
48:20 +0100 Cc: [EMAIL PROTECTED] Subject: Re: rc & runsystem I think you mean the files in /etc/login, and not /etc/{ttys,motd}. config/ChangeLog 2004-12-16 Alfred M. Szmidt <[EMAIL PROTECTED]> * Makefile ($(installed_logins)): Don't overwrite already existing

Re: rc & runsystem

2004-12-17 Thread Alfred M. Szmidt
It's still not a configuration file. You are wrong. ___ Bug-hurd mailing list [EMAIL PROTECTED] http://lists.gnu.org/mailman/listinfo/bug-hurd

Re: rc & runsystem

2004-12-17 Thread Alfred M. Szmidt
Was there anything wrong with this patch? ___ Bug-hurd mailing list [EMAIL PROTECTED] http://lists.gnu.org/mailman/listinfo/bug-hurd

Re: rc & runsystem

2004-12-17 Thread Roland McGrath
>> Or did you mean "not to overwrite existing stuff"? If you didn't >> then we both agree, and this whole discussion was pointless. > >No, I said overwrite selectively. > > No you didn't. Give me a fucking break. When you find yourself arguing with someone about what he himself s

Re: rc & runsystem

2004-12-17 Thread Alfred M. Szmidt
> Or did you mean "not to overwrite existing stuff"? If you didn't > then we both agree, and this whole discussion was pointless. No, I said overwrite selectively. No you didn't. You said: , | If you want to use make install for _overwriting_existing_stuff_ | selectively, you shou

Re: rc & runsystem

2004-12-17 Thread Roland McGrath
> Or did you mean "not to overwrite existing stuff"? If you didn't then > we both agree, and this whole discussion was pointless. No, I said overwrite selectively. You want it to overwrite some executables and not others, just because these two are implemented as scripts you don't use the vanill

Re: rc & runsystem

2004-12-17 Thread Alfred M. Szmidt
> This is getting silly, maybe I should just ask for that > branch... :-) Making a stink because you don't want to type --disable-install-runsystem in configure commands when doing your own hacking is what's silly. It was a joke, lighten up... The vanilla use of make install is

Re: rc & runsystem

2004-12-17 Thread Roland McGrath
> This is getting silly, maybe I should just ask for that branch... :-) Making a stink because you don't want to type --disable-install-runsystem in configure commands when doing your own hacking is what's silly. The vanilla use of make install is to populate an empty $(prefix) directory with a p

Re: rc & runsystem

2004-12-17 Thread Roland McGrath
> I agree with Roland that `make install' is not a package manager nor > should we pretend that it is. On the other hand, the files we are > talking about are basically templates that a package manager would > modify and if we consider them as templates then I think not > installing them is fine.

Re: rc & runsystem

2004-12-17 Thread Neal H. Walfield
> It is about what _users_ expect, and this is what users > expect, you have Neal, Harley and Marco back over there who think this > is a good. My position is the following: it is annoying to have which I have modified overwritten. As I understood the original patch, it elided this annoying beha

Re: rc & runsystem

2004-12-17 Thread Alfred M. Szmidt
I have already told you. make install is there to install the files that the sources provide. Fine, how about removing rc, motd, ttys, etc completely from the sources, and not install them at all? And you are contradicting yourself anyway, since some files don't get overwriten (the files p

Re: rc & runsystem

2004-12-17 Thread Roland McGrath
I have already told you. make install is there to install the files that the sources provide. ___ Bug-hurd mailing list [EMAIL PROTECTED] http://lists.gnu.org/mailman/listinfo/bug-hurd

Re: rc & runsystem

2004-12-17 Thread Alfred M. Szmidt
Sorry, you are just wrong. Bah, really, can't you come up with anything better then "Because"? I'm not stupid to reject a _good_ reason, but you don't have any reason at all! Atleast be a good enough sport to give one reason why it is "wrong". And if it is wrong, then the whole override_conf h

Re: rc & runsystem

2004-12-17 Thread Roland McGrath
Sorry, you are just wrong. ___ Bug-hurd mailing list [EMAIL PROTECTED] http://lists.gnu.org/mailman/listinfo/bug-hurd

Re: rc & runsystem

2004-12-17 Thread Alfred M. Szmidt
It isn't about me or my hacks, you said yourself I could get a branch for that. It is about what _users_ expect, and this is what users expect, you have Neal, Harley and Marco back over there who think this is a good. ^ Good Thing (TM). __

Re: rc & runsystem

2004-12-17 Thread Alfred M. Szmidt
make install is make install. It is not a package manager. Get used to it. No, I won't get used to it since the behaviour of `make install' is wrong. That _I_ use it as a package manager is totaly irrelevant to the dicussion. Clobbering files that *will and should* get modifed by users i

Re: rc & runsystem

2004-12-17 Thread Roland McGrath
make install is make install. It is not a package manager. Get used to it. If you would like a configure switch or makefile variable to enable some nonstandard behavior for your convenience while hacking, that is fine by me. But don't get any fantasies about what the proper behavior is. ___

Re: rc & runsystem

2004-12-17 Thread Alfred M. Szmidt
It has always been normal for packages to clobber config files in make install. Not in my experience, it has been normal to just not install configuration files at all and let the user do it manually. I would actually file a bug report for _anything_ that clobbers configuration files. Infa

Re: rc & runsystem

2004-12-16 Thread Roland McGrath
It has always been normal for packages to clobber config files in make install. That's why users don't do make install, but use package managers. These are not really config files in the normal sense that users generally edit them. They are things in the GNU package that packagers usually change.

Re: rc & runsystem

2004-12-16 Thread Harley D. Eades III
"Alfred M. Szmidt" <[EMAIL PROTECTED]> writes: > daemons/ChangeLog > 2004-12-16 Alfred M. Szmidt <[EMAIL PROTECTED]> > > * Makefile (targets): Removed `rc' and `runsystem'. > (special-targets): Variable removed. > (runsystem, rc): Targets removed. > (FORCE, $

rc & runsystem

2004-12-16 Thread Alfred M. Szmidt
c and runsystem as configuration files, and overwriting config files is just a no-no in my book (and I can't really count how many times I got seriously pissed because it munged my tweaked versions). Please? This is like the third time I'm annoying everyone with rc/runsystem. :-) dae

Re: rc & runsystem

2004-12-16 Thread Alfred M. Szmidt
ts = rc getty mail.local console-run runttys runsystem -special-targets = rc runsystem +targets = getty mail.local console-run runttys SRCS = rc.sh getty.c lmail.c console-run.c runttys.c installationdir = $(libexecdir) @@ -33,7 +32,6 @@ include ../Makeconf -rc: rc.sh getty: getty.o ../libsh

Re: rc & runsystem

2004-12-16 Thread Marco Gerards
"Alfred M. Szmidt" <[EMAIL PROTECTED]> writes: >It's quite annoying, I agree. The same is true for /etc/motd and >/etc/ttys, IMHO. > > Comments in the code contradict you... > > [hurd]/config/Makefile > > # Files that are copied verbatim to $(sysconfdir). But we never want > # to overwri

Re: rc & runsystem

2004-12-16 Thread Neal H. Walfield
> -# Copyright (C) 1996,99 Free Software Foundation, Inc. > +# Copyright (C) 1996,99,04 Free Software Foundation, Inc. This copyright year it definitely wrong. I have no objections to the rest of the patch. ___ Bug-hurd mailing list [EMAIL PROTEC

Re: rc & runsystem

2004-12-16 Thread Alfred M. Szmidt
I think you mean the files in /etc/login, and not /etc/{ttys,motd}. config/ChangeLog 2004-12-16 Alfred M. Szmidt <[EMAIL PROTECTED]> * Makefile ($(installed_logins)): Don't overwrite already existing login files unless asked for. Index: config/Makefile =

Re: rc & runsystem

2004-12-16 Thread Marco Gerards
"Alfred M. Szmidt" <[EMAIL PROTECTED]> writes: > If you are making a system you should tweak both of them anyway, and > in a sub-hurd it doesn't matter, since you'll be punted into a > single-luser shell (or you should) by init. I personally consider rc > and runsystem as configuration files, and

Re: rc & runsystem

2004-12-16 Thread Alfred M. Szmidt
# If override_conf is set to `t' then install even on top of existing # files. Didn't know about this, cool. Infact, I wouldn't mind such a solution for rc/runsystem. ___ Bug-hurd mailing list [EMAIL PROTECTED] http://lists.gnu.org/mailman/listinfo/bug-hurd