Re: Continuously-built Nix-based QEMU image

Am Dienstag, 8. November 2011, 02:00:15 schrieb Samuel Thibault: > Arne Babenhauserheide, le Tue 08 Nov 2011 01:15:05 +0100, a écrit : > > But “Assertion `retry == FS_RETRY_NORMAL` failed” > > Yes, because the automatic build hasn't reached the latest revision > which does work. Ah, OK :) Definit

Re: Continuously-built Nix-based QEMU image

Arne Babenhauserheide, le Tue 08 Nov 2011 01:15:05 +0100, a écrit : > But “Assertion `retry == FS_RETRY_NORMAL` failed” Yes, because the automatic build hasn't reached the latest revision which does work. Samuel

Re: Continuously-built Nix-based QEMU image

Am Dienstag, 8. November 2011, 00:24:03 schrieb Ludovic Courtès: > Hello! > > After a fair amount of time debugging and discussing with Samuel and > others, here’s a Nix[*]-based, cross-built GNU/Hurd QEMU image: > > http://hydra.nixos.org/jobset/gnu/hurd-master That sounds great! But “Assertio

Continuously-built Nix-based QEMU image

Hello! After a fair amount of time debugging and discussing with Samuel and others, here’s a Nix[*]-based, cross-built GNU/Hurd QEMU image: http://hydra.nixos.org/jobset/gnu/hurd-master This image is continuously built off the Hurd and Mach VCS, and based on a libc & libpthread snapshot (the u

Re: [PATCH] Provide modern Autoconf initialization

Hi! In GNU Mach, we have a version.m4 where these fields are defined one and for all. (I argue that a package's version definition does not really belong into a generic, otherwise version-agnostic configuration system's file.) (But on the other hand, *version*.m4 is likely not the most suitable

Re: [PATCH] Provide modern Autoconf initialization

Samuel Thibault skribis: > Ludovic Courtès, le Mon 07 Nov 2011 22:41:02 +0100, a écrit : >> OK to apply? :-) > > Mmm, 0.3 is in Makeconf too, ideally it'd be in one place only. I would mv Makeconf{,.in} and then substitute PACKAGE_VERSION in it. WDYT? (Then again, deriving shared library names

Re: [PATCH] Have `rc' and `runsystem' use /bin/sh

Samuel Thibault skribis: > Ludovic Courtès, le Mon 07 Nov 2011 22:43:22 +0100, a écrit : >> These two scripts really need a plain POSIX shell, not necessarily >> Bash. This patch changes the shebang accordingly. > > Mmm, are "local" and functions really in POSIX shells? dash refuses > the functi

A tale of bootstrap ld.so debugging

Hello! While cross-building a GNU/Hurd QEMU image, I stumbled upon a series of bugs, the last one of which led to a one-liner adding libgcc_s.so to the image. Lack of libgcc_s.so prevented the initial /hurd/exec to run. The kernel debugger showed a backtrace in ld.so hinting at an undefined symb

Re: [PATCH] Have `rc' and `runsystem' use /bin/sh

Ludovic Courtès, le Mon 07 Nov 2011 22:43:22 +0100, a écrit : > These two scripts really need a plain POSIX shell, not necessarily > Bash. This patch changes the shebang accordingly. Mmm, are "local" and functions really in POSIX shells? dash refuses the function keyword, but apparently f() {

Re: [PATCH] Provide modern Autoconf initialization

Ludovic Courtès, le Mon 07 Nov 2011 22:41:02 +0100, a écrit : > OK to apply? :-) Mmm, 0.3 is in Makeconf too, ideally it'd be in one place only. Samuel

Re: [PATCH] Allow /hurd/init to be a symlink

Ludovic Courtès, le Mon 07 Nov 2011 22:15:43 +0100, a écrit : > Yes (though it’s the administrator who set up the symlink, so it could > be argued that it can be trusted to not be infinite.) Yep, but it's probably better to get an ELOOP error than an stack overflow. > How about this updated versi

[PATCH] Have `rc' and `runsystem' use /bin/sh

Hello, These two scripts really need a plain POSIX shell, not necessarily Bash. This patch changes the shebang accordingly. (Tested with Bash 4.1 run as /bin/sh.) OK? Ludo’. diff --git a/daemons/rc.sh b/daemons/rc.sh index 5778094..e9f7fca 100644 --- a/daemons/rc.sh +++ b/daemons/rc.sh @@ -1,

[PATCH] Provide modern Autoconf initialization

Hello, OK to apply? :-) Ludo’. diff --git a/configure.in b/configure.in index 635242d..05b959b 100644 --- a/configure.in +++ b/configure.in @@ -1,7 +1,7 @@ dnl Process this file with autoconf to produce a configure script. AC_REVISION([$Id: configure.in,v 1.38 2008/11/17 11:34:18 tschwinge Ex

Re: [PATCH] Allow /hurd/init to be a symlink

Hi, Samuel Thibault skribis: > Ludovic Courtès, le Mon 07 Nov 2011 21:35:06 +0100, a écrit : >> + lookup_init: >>err = dir_lookup (root_pt, initname, O_READ, 0, >> &retry, pathbuf, &startup_pt); >>if (err) >> @@ -232,6 +233,13 @@ diskfs_start_bootstrap () >>free

Re: [PATCH] Allow /hurd/init to be a symlink

Ludovic Courtès, le Mon 07 Nov 2011 21:35:06 +0100, a écrit : > + lookup_init: >err = dir_lookup (root_pt, initname, O_READ, 0, > &retry, pathbuf, &startup_pt); >if (err) > @@ -232,6 +233,13 @@ diskfs_start_bootstrap () >free (exec_argv); >assert_perror (er

[PATCH] Allow /hurd/init to be a symlink

Hello! The attached patch allows /hurd or /hurd/init to be a symlink to an absolute path. OK to apply? Thanks, Ludo’. diff --git a/libdiskfs/boot-start.c b/libdiskfs/boot-start.c index e73e4d3..30b54d3 100644 --- a/libdiskfs/boot-start.c +++ b/libdiskfs/boot-start.c @@ -1,5 +1,5 @@ /* - Copy