IF YOU WOULD LIKE TO GET AN ACCOUNT, please write an
email to Administrator. User accounts are meant only to access repo
and report issues and/or generate pull requests.
This is a purpose-specific Git hosting for
BaseALT
projects. Thank you for your understanding!
Только зарегистрированные пользователи имеют доступ к сервису!
Для получения аккаунта, обратитесь к администратору.
See http://www.opennet.ru/openforum/vsluhforumID3/86239.html#1
for a query that has led to this one; in particular,
- xdm dropped (won't log in root and there are no users yet);
- network is brought up and configured via DHCP by default;
- apt-get works out-of-box;
- default image size is twice the chroot size.
This isn't ready for general consumption (just as centos one)
but the notion of REPO is floating around along with apt-conf
thoughts, and it might still be useful to someone poking around
conf.d/test.mk.
Request hasher-pkg-init.spec from mike@ or led@ if interested;
the experiments were carried out using openSUSE 11.4 repository
and slightly patched hasher (cpio blacklist for devices).
That sub/stage2/install2 was somewhat clumsy actually as it looked
like a hierarchical thing while being a substitution thing:
generic stage2 would get put in place renamed as install2.
This could only get worse with hierarchical features which have
already been both requested and considered for quite a time,
and "stage2 at install2" reads much more naturally.
There were heaps of "if type -t git" there already;
it wasn't an unintentional mishap but rather a moderate
copy-paste to get the use cases, and now these seem to
have essentially settled.
So time to scrap some dups.
NB: the scripts in the generated profile can't rely on
the contents of the metaprofile (these need to be able
to work in standalone case either), so a bit of crap
still lurks there.
The prerequisites for a cleanup after a successful build
were somewhat weird at this point; now the rules are:
- if DEBUG level is more than 1 or CHECK is set, don't do it;
- otherwise if at least one of the following conditions is true:
+ there's more than one target being built in a row;
+ the build was run by e.g. alterator-mkimage;
+ metaprofile directory is read only
...then do a distclean.
If these are still weird or feel unsuitable for profile hacking,
drop me a note (or a patch).
Actually the templates pretending to be usable missed the whole
interactivesystem (sysvinit would get pulled in by services as well).
Fixed somewhat but time and practice will tell.
mkimage and hasher can make use of qemu to run
non-native binaries while working on the chroots;
thanks kas@, manowar@ and sbolshakov@ for implementing
this functionality as well as providing nice examples
through mkimage-profiles-arm and mkimage-profile-armrootfs.
This required the architecture check to be added since baking
a tarball with "arm" as its specified arch and x86_64 inside
isn't particularly good thing to let slip through; however
the implementation is quite fragile, bugreports and patches
are seriously welcome.
NB: APTCONF evaluation order between lazy make and nimble shell
turned out to be quite a delicate issue in this particular case.
The only thing to be fixed was setarch(8) symlinks assumption
that is correct for x86 but not for ARM.
There's also some hasher(7) setup to be done:
mkdir -p ~/.hasher
echo >> ~/.hasher/config <<-EOF
def_target=arm
#cache_dir=$HOME/tmp # depends on RAM/storage configuration
EOF
...and of course apt(8) should be properly set up too.
An example PoC build on a CM-A510 board (tmpfs):
$ make BRANDING=altlinux-centaurus ve/bare.tar.gz
** ARCH: arm
18:10:45 initializing BUILDDIR: build/
18:10:45 preparing distro config: build/distcfg.mk
18:10:46 starting image build: tail -f build/build.log
18:14:49 done (4:02)
** image: $TMP/out/bare-20120706-arm.tar.gz [23M]
There's no need to repeat the typical openssh-* triade
all over the place; those who need server and client
are better off pulling in "openssh" pkglist, and those
needing a particular package should specify it.
Yes, mkimage-profiles is now able to build VM disk images.
So far the support is pretty basic:
- a single hard drive image with a single partition/FS
- only stock root password is configurable
- LILO is hardwired as a bootloader
The resulting images tend to boot under qemu/kvm though.
Please see doc/vm.txt for the warning regarding additional
privileges and setup required. This was started back in
February but I still hoped to avoid sudo/privileged helper
(and libguestfs is almost as undistributable as can be)...
Thanks:
- http://blog.quinthar.com/2008/07/building-1gb-bootable-qemu-image-using.html
- Alexey Morarash who reworked that as https://github.com/tuxofil/linsygen
- led@, legion@, vitty@, aen@ for providing advice and inspiration
It somehow managed to evade me that $(TMP) might be uninitialized;
definitely should be checked before stuffing into sed substitution
command.
NB: this could be done in pure make but my take was less readable.
Thanks shadowsbrother/gmail for hitting and reporting this.
Just like livecd-install, graphical installer KMS support
looks better as an optional part of install2 feature.
Of course it's optional only if the release manager is fine
with VESA drivers and not KMS-requiring intel/radeon/nouveau;
thanks led@ for a confirmation just in case.
This further refines the modular build by making
metadata being a clearly separated feature rather
than having to rely on runtime tests, and also by
moving the code which cares for kernel bits of base
installation (.base list) in a feature of its own.
There's more to it but let's get the ball rolling first.
It's hard to tell a successful build from a failed one
if downstream hides the exit code; it's useless to continue
a `for' loop if a pipe shoves that to a subshell; well it seems
that a bashism is worth a thousand quirks with extra fds here.
Minor regexp enhancements are also due.
reports.mk made a bit more resilient/prudent either.
Sometimes it's desirable to provide the kernel supporting
maximal amount of RAM on the system; bad news is that x86
has a kludge named PAE, good news is that x86_64 doesn't
need it at all; but now we must be able to choose between
those.
BIGRAM will hold the flavour needed.
Initial SPICE support has been added for kvm/libvirt installation
and boot-up using qxl and spice by default as proposed by shaba@.
VirtualBox part is shifted a level deeper correspondingly
but otherwise stays the same.
A pretty common issue breaking the image build is inter-package
file conflict resulting in hsh-install failure down there.
Let's bring that back to attention conveniently.
There is at least one known deficiency for mkimage-profiles:
build.log will be truncated if verbose mode is enabled and
hasher version is lower than 1.3.22.
The check is done here since it's where the logging is arranged,
and doing it in image.in/Makefile would result in the warning
about log-truncating software being truncated by the said software.
Thanks Max Kosmach for reporting this inobviousity.
The output was still somewhat ragged in 80x24 terminal window
with fmt(1) which wasn't anticipating the word length difference
subsequent column(1) would have to cope with later on.
Thanks Loic Cattani for his shell columnizer implementation:
https://github.com/Arko/Columnize
As was noted by Alexey Shabalin in libosinfo context,
current ALT Linux images tend to lack ISO9660 metadata
-- which they did have back in the day of Master 2.4.
Please note that the data collection occurs this way
due to mkimage's config.mk resetting the values to be
empty; this was worked around by using another config
file, $(BUILDDIR)lib/iso.mk, and including it later
but that would require a separate target with per-target
CONFIG variable which isn't elegant at all given the need
to actually build up the metadata set.
So the variables were changed (to be more readable anyways)
and then proxied back to BOOT_*. This might be cleaned up
some day after the inclusion order is tweaked or mkimage
defaults get set-if-unset-yet (?=).
openssh-server is in need indeed on almost any server instance;
thanks Aleksey Cheusov for reporting the shortage.
This might be amended in the future but is reasonable right now.
As noted in doc/assumptions.txt, the SHELL based target tracing
only works for rules with recipes, even empty but present ones.
The simplest thing to do is hooking "; @:" onto the rule's tail
(one-liner with a non-printing shell builting "true" command).
The purpose is being able to examine particular target interdependency
graph for a given image having been configured to avoid convoluted
dependencies (loops in particular).
The implementation is based on SHELL hook hint by John Graham-Cumming:
http://cmcrossroads.com/ask-mr-make/6535-tracing-rule-execution-in-gnu-make
As was duly noted by Leo-sp50, both server.mk and desktop.mk
duplicate a few bits layered over bare distro/installer which
happened to be both a dependency (thus should reduce redundancy)
and a "real distro" target (well, it doesn't just work yet, need
to provide networking and sources.list in install2 by hand).
Fixed by moving a "node" to distro/.installer along with typical
additions and leaving a bare installer as is by now; there's a
need to get it working at least for DHCP/ftp.altlinux.org case.
A minimal chroot supporting extension via apt-get;
vitals if built on Sisyphus as of Jan 16, 2012:
i586: 13M tar.xz, 58M chroot (33M w/o /usr/share/{doc,locale,man})
x86_64: 14M tar.xz, 60M chroot (35M w/o /usr/share/{doc,locale,man})
Trivial fixups (extra checks) added to two script hooks.
As was found out by Vladimir Karpinsky (thanks for patience!),
the autochosen directory might still have too restrictive mount
options -- nodev and/or noexec. Hopefully the diags are a bit
better and faster by now.
It was briefly mentioned in QUICKSTART but somehow managed
to evade the commandlines provided. And while at it, let's
make errors like this more explicit to avoid extra lookups.
Oh, and fix QUICKSTART so that readers miss the hassle. :)
Thanks Vladimir Karpinsky for pointing this problem out.
doc/variables.txt was missing the already-existing BUILDLOG
variable description, and ARCHES got added during multi-target
toplevel rewrite. Other minor fixes come as appropriate.
The former toplevel Makefile is now toplevel main.mk;
this change allows for multi-target, multi-arch processing
in the current toplevel Makefile.
As the "build" symlink semantics change quite considerably
when one is doing bulk builds (several pruned builddirs might
be useful for comparison), BUILDDIR is now much more likely
to be recreated: the cases when it will persist are when it's
either a single-image build or when the prefix hasn't changed.
There are some more or less subtle bugfixes and enhancements
all over the map as well.
Done within 20111230..20120102 timeframe, actually...
First, let's not do rsync --delete on an unverified target dir
again: the lesson was learned during a subway hacking session
and I must say that SSDs are frightening fast (even if it was
more than a second to realize what happens and terminate the
extermination before it got /home, thanks xterm).
Second, let's use a variable for common name and make's own
realpath function instead of external binary.
Initial openSUSE package base taming effort has shown that
relatively few things should be fixed; subst has been generalized
as -i option to sed(1) since its introduction, so let's just fix it.
The idea is to check:
- the reachability of every target
used to build the image in question;
- the availability of all the package lists
and subsequently packages for that image;
- the lack of "dangling" intermediate targets,
features, pkglists, hooks etc.
So far only the first step is implemented --
it's easy and somewhat helpful already for
make CHECK=1 all
distro/.base target used to pull in localboot syslinux config
snippet which might be too early for some of the further distros;
it's a quite fragile equilibrium which was shifted a bit by imz@
(see #26606). Feel free to reopen the discussion though, things
might be tweaked so that localboot might be desirable on almost
every image even if with lower priority...
As was (quite reasonably) asked by someone and me too,
why should a successful build yield a *red* line
(a grep's default)?
So now it's new and improved, 25% free and so forth:
with a successful build you get a green line, while
errors from a broke one result in red ones.
Clinically tested in both b/w and w/b colour schemes;
in case you're not satisfied, please return original
ANSI_OK and ANSI_FAIL values to the colour dealer
and pass your favourite ones instead.
The features might get copy-pasted (or even copied-and-pruned)
when initialized; there's an unneccessary duplication of the
function name in the line adding it to FEATURES list, thus
prone to being forgotten and causing some havoc later on.
It was wrong in the first place but tackling this with some
double-colon rules ran into terminality issues, and further
tortures were considered unneccessary.
The current solution isn't perfect (no completely transparent
function name registration upon corresponding target being called)
but at least it is an improvement...
