gmane.linux.sourcemage.discuss

Re: Spell Deprecation

Sukneet Basuta — 2012-05-25T23:22:25

Thanks! I couldn't seem to find that when I searched the wiki.

Re: Spell Deprecation

Paul Vinogradov — 2012-05-25T22:04:01

http://wiki.sourcemage.org/Spell_Deprecation

I don't think anything have been changed much here since then...

В Fri, 25 May 2012 17:51:16 -0400
Sukneet Basuta <sukneet< at >gmail.com> пишет:

Spell Deprecation

Sukneet Basuta — 2012-05-25T21:51:16

Can someone please outline the proper procedure to deprecate spells
(and possibly post it on the wiki)?

remove the spell if there is no direct replacement for it.
If there is, you add an up_trigger to cast its replacement and an
up_trigger to dispel_self, replace the contents of PRE_BUILD, BUILD,
and INSTALL with true, comment out the SOURCE info in DETAILS, and
remove any remaining files (other than HISTORY). Is that correct?

In case anyone is wondering, I want to deprecate howl (it currently
fails to build and is long dead) and replace any current depends on it
with -sub COMPAT_HOWL avahi.

Re: SPELL > SCRIPT

Jaka Kranjc — 2012-05-24T09:48:04

The regex would have to be improved though, since there are also "legitimate" 
uses of the bracy form around, for example when the variable name is followed 
by any [[:alpha:]_]. I doubt there's many in spell files, if any at all though, 
so a manual inspection of everything that isn't followed by a slash may be 
enough.

Does it make sense? Both are just styling options, so I have no real argument 
for or against either.

LP

Re: SPELL > SCRIPT

flux — 2012-05-24T09:50:33

Thomas Orgis (thomas-forum< at >orgis.org) wrote [12.05.24 18:33]:

In either case they're both wrong, because they need to be quoted.
Beyond the quoting, the braces are only going to matter if the variable
name needs to be disambiguously differentiated from what follows.

Re: SPELL > SCRIPT

Thomas Orgis — 2012-05-24T09:33:03

Am Thu, 24 May 2012 21:19:10 +1200
schrieb AbH Belxjander Draconis Serechai <belxjander< at >gmail.com>: 


Something along that, yes. Though, naturally, I'd do something along

perl -pi -e 's:\$\{SCRIPT_DIRECTORY\}:\${SPELL_DIRECTORY}:g; s:\$SCRIPT_DIRECTORY:\$SPELL_DIRECTORY:g;'  $spell_file

others would use sed ... but, well, that is not the point: The point is whether we should start an argument about ${varname} vs. $varname;-)


Alrighty then,

Thomas
_______________________________________________
SM-Discuss mailing list
SM-Discuss< at >lists.ibiblio.org
http://lists.ibiblio.org/mailman/listinfo/sm-discuss

Re: SPELL > SCRIPT

Thomas Orgis — 2012-05-24T09:15:19

Am Thu, 24 May 2012 10:04:04 +0200
schrieb Jaka Kranjc <smgl< at >lynxlynx.info>: 


What about (automated) purging of $SCRIPT_DIRECTORY and deprecation of it, then?


Alrighty then,

Thomas
_______________________________________________
SM-Discuss mailing list
SM-Discuss< at >lists.ibiblio.org
http://lists.ibiblio.org/mailman/listinfo/sm-discuss

Re: SPELL > SCRIPT

Jaka Kranjc — 2012-05-24T08:04:04

The division predates even my involvement, so it must be ancient history. I 
agree that SPELL_DIRECTORY should be prefered in new code though, since it is 
more precise.

LP

Re: SPELL > SCRIPT

Thomas Orgis — 2012-05-24T08:00:04

Am Wed, 23 May 2012 01:13:28 +0400
schrieb Vlad Glagolev <stealth< at >sourcemage.org>: 


Oh, is SCRIPT the old one and SPELL the new name? What's the history of these two?


Alrighty then,

Thomas
_______________________________________________
SM-Discuss mailing list
SM-Discuss< at >lists.ibiblio.org
http://lists.ibiblio.org/mailman/listinfo/sm-discuss

Re: Announcing stable-rc-0.61 (20120517) chroot image release

Vlad Glagolev — 2012-05-22T21:47:48

On Wed, 23 May 2012 09:41:47 +1200
AbH Belxjander Draconis Serechai <belxjander< at >gmail.com> wrote:


that's an interesting idea, but I'm not sure all the spells are pretty
ready for cross-compilation.

though I track T2 project [0], which has _A LOT_ of work on this done.
And it's aimed on providing clear packages built from source without
any extra patches, only those needed for cross-compile.

[0] http://www.t2-project.org/about.html

Re: Announcing stable-rc-0.61 (20120517) chroot image release

Vlad Glagolev — 2012-05-22T21:36:45

then you seem misunderstanding me for the 3rd time as I do for you :)


I didn't mean the images themselves, but theirs contents.
it can be even just a spell-list taken from it.


i686


I don't think so, it's "Intel Pentium Pro and Clones" in sorcery menu,
which doesn't have such spec in it.


nothing extra'ish


Of course I didn't want to mean that :) The idea is to use it by us
while preparing an ISO. Kernel's config, spell-list, default configs
in /etc and so on.

SPELL > SCRIPT

Vlad Glagolev — 2012-05-22T21:13:28

For all the spellwriters,

even while $SCRIPT_DIRECTORY and $SPELL_DIRECTORY are equal in
functionality, please use the last one as it actually makes more sense.

Re: [SM-Announce] Announcing stable-rc-0.61 (20120517) chroot image release

Sukneet Basuta — 2012-05-18T04:31:44

Just the GF116? or GF11x chipsets? If its the former, that includes
the GT 545 and GTX 550.


Fair enough. A simple warning message may suffice however.


In the bug report, NoName replied:
"I confirm that the livecd kernel 3.1.0-7.fc16 works correctly with the livecd
and an installed fc16. All other kernels that i tried (kernels 3.0.x and
kernels 3.2.x upwards) would result in the same problem"

That seems to suggest to me that he/she used the kernel with the
nouveau driver. Its not explicitly stated though, so maybe I'm wrong.

Re: Announcing stable-rc-0.61 (20120517) chroot imagerelease

flux — 2012-05-18T00:26:03

Vlad Glagolev (stealth< at >sourcemage.org) wrote [12.05.18 08:34]:

No, they cannot be used for producing ISOs, as I explicitly stated.
There are various reasons for this, some of which I explicitly pointed
out in my previous mail.


Because you haven't tried building one yet. :)

An ISO consists of two parts: the ISO runtime itself (what runs when you
boot it), and the "payload" that gets installed to the target
(this is system.tar.bz2 on 0.10). The images you're creating cannot be
directly used for either the payload or the runtime. If you are
interested in producing either payloads or runtimes (or both), please
use the cauldron scripts.


Generic is meaningless for actual production. i486 is meaningful. Are
these i486? i586? ...? Did you build them using --march=native? What
were the CFLAGS in use?

These images are certainly helpful, but if you are serious having them
used for official installs, then you should be much more specific and
explicit about them.


That _is_ the process (part of it). You cannot su

Re: Announcing stable-rc-0.61 (20120517) chroot image release

Vlad Glagolev — 2012-05-17T23:34:40

noone told they're. but theirs contents can be used for producing ISO.
the contents -- is just spells.


I just can't imagine any other way of reusing chroot contents for
building an ISO :)


both generic. in my 1st email a week ago.


yes. I'm just talking about contents of chroot images for ISO-creation,
not about the process itself.

Re: [SM-Announce] Announcing stable-rc-0.61 (20120517) chroot image release

Arjan Bouter — 2012-05-17T22:48:39

On Thu, 17 May 2012 18:31:22 -0400
Sukneet Basuta <sukneet< at >gmail.com> wrote:



That's why I mentioned the gtx 550 card specifically. The bug is valid for
every card with the same chipset on it, which are probably more cards.

While I agree it seems silly, it might make more sense when you look at it this
way:
It's quite easy to re-enable it if you know your card is safe, but if it's your
only box you would have no chance of searching for the solution.



That's common, for a livecd to work. Quite a few use the basic vesa driver
instead of card-specific drivers to maximize the amount of supported cards
while minimizing the amount of drivers on the livecd.

to save you some time searching for examples: opensuse's livecd works (vesa
driver), linux mint 12 and their debian edition crash hard because they use
nouveau by default.

Arjan Bouter

+===================
Source Mage GNU/Linux developer,
http://www.sourcemage.org
+============

Re: [SM-Announce] Announcing stable-rc-0.61 (20120517) chroot image release

Sukneet Basuta — 2012-05-17T22:31:22

I think you are overreacting. We don't know if every Nvidia card is
affected. It seems silly to me to prevent users from using nouveau if
it works with their card.

Also the redhat bug report you linked to stated that the kernel on
their livecd (3.1.0-7.fc16) worked. Using a stable kernel seems like a
better idea than using mainline for the stable grimoire.

Re: [SM-Announce] Announcing stable-rc-0.61 (20120517) chroot image release

Arjan Bouter — 2012-05-17T22:10:35

On Fri, 18 May 2012 01:49:07 +0400
Vlad Glagolev <stealth< at >sourcemage.org> wrote:


I agree the kernel is important, obviously, but it won't leave a good
impression if our stable tarball's kernel locks up on specific hw.


That won't work unless that kernel has nomodeset in CONFIG_CMDLINE or
nouveau support is disabled completely, in which case the kernel will fall back
to vesa.
On new nvidia cards the kernel will lock up as soon as the nouveau module is
loaded.

I suggest we disable nouveau in the kernel for the chroot until the default
kernel version is > 3.4.x so we at least have the option of giving people a
warning.

Arjan Bouter

+===================
Source Mage GNU/Linux developer,
http://www.sourcemage.org
+============

Re: Announcing stable-rc-0.61 (20120517) chroot imagerelease

flux — 2012-05-17T22:03:51

Vlad Glagolev (stealth< at >sourcemage.org) wrote [12.05.18 06:54]:

As you've not experienced producing an ISO, I'll have to simply disagree
and say "no". These are not ISO images. They don't boot, but they could
possibly be used as possession images. They cannot be used for producing
ISOs, as they include way too much.

They only way they could be used in producing a new ISO is by using the
prebuilt spells from these tarballs for preparing the ISO runtime.
However, as these tarballs were produced outside the ISO preparation
process and you haven't published the exact process by which you
produced them, there is no way to know what options the spells were
built with, for what specific CPU/architecture, etc.

If you want to build chroots for the purpose of ISO preparation, please
use the Cauldron scripts to do so. That's why they exist.

Of course I'm not implying these images are worthless. They can be quite
helpful in bootstrapping new systems via a possess-like method -- boot
an existing ISO, whether SMGL or o

Re: Announcing stable-rc-0.61 (20120517) chroot image release

Vlad Glagolev — 2012-05-17T21:54:23

These images cannot boot themselves.
Therefore they cannot be used as replacements for an ISO.

excluding philosophical chit-chat, new installation ISO might be based
on them + syslinux + required scripts (which are the part of Cauldron)
for step-by-step installation process. Just try not to focus on a
"based" word so much.

Re: Announcing stable-rc-0.61 (20120517) chroot imagerelease

flux — 2012-05-17T21:23:43

Vlad Glagolev (stealth< at >sourcemage.org) wrote [12.05.18 05:38]:
<snip>
</snip>

These images are not basesystem chroots, they are full chroots.
Therefore they cannot be used to prepare an ISO. They would possibly be
replacements for an ISO (hopefully with sufficient testing, etc.).