[gentoo-dev] USE="minimal" for kernel sources

[gentoo-dev] USE="minimal" for kernel sources

[Petteri Räty]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

I have a couple of old machines I maintain and emerging and unmerging
kernel sources take a while because there are so many files. Also one
set of gentoo sources takes about 230MB of disk space. By removing stuff
not belonging to x86 I was able to succesfully run make with 58MB/230MB
removed. The stuff I removed:
arch/* except i386 and x86_64
include/asm-* expect asm-generic, asm-i386 and asm-x86_64

So I propose we implement a minimal USE flag in the kernel-2 eclass that
would make the cleaning I succested before the merging. One problem
coming from the clean is that make clean does not work and at the moment
it is run before unmerging, which is of course a good thing. If the
kernel devs think this is a good idea, I can make an implementation for
this.

Regards,
Petteri Räty (Betelgeuse)
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFDHJfDcxLzpIGCsLQRArvaAJ44MvDquwIAhq35ESwBAqyQZXcZoACglatc
A/ecjOeVr70k14T7rQpzu9U=
=JctU
-----END PGP SIGNATURE-----
-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] USE="minimal" for kernel sources

[Michael Hanselmann]

[-- Attachment #1: Type: text/plain, Size: 498 bytes --]

Hello

> The stuff I removed:
> arch/* except i386 and x86_64
> include/asm-* expect asm-generic, asm-i386 and asm-x86_64

> So I propose we implement a minimal USE flag in the kernel-2 eclass that
> would make the cleaning [...|

The idea isn't too bad, but if you implement it, please do it in a cross
plattform compatible way, not x86-only.

Greets,
Michael

-- 
Gentoo Linux Developer using m0n0wall | http://hansmi.ch/
Tomorrow's computers some time next month.
    	-- DEC

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] USE="minimal" for kernel sources

[Petteri Räty]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Michael Hanselmann wrote:
> Hello
> 
> The idea isn't too bad, but if you implement it, please do it in a cross
> plattform compatible way, not x86-only.

Well I didn't explain the implementation, but of course I would myself
only add the implementation for x86. As I noticed while testing for this
i386 needs stuff from x86_64 so there could be similar situations with
other arches. Because of that reason I would leave it to arch teams to
add their own list of needed directories.

> 
> Greets,
> Michael
> 

Regards,
Petteri Räty (Betelgeuse)
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFDHKi3cxLzpIGCsLQRAh2tAJ9EjScnpva64gPevwbU3L8/gmy89QCfWw8H
2Tw3FSb98yrZWGxwPxDD0xs=
=pHLw
-----END PGP SIGNATURE-----
-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] USE="minimal" for kernel sources

[solar]

[-- Attachment #1: Type: text/plain, Size: 1344 bytes --]

On Mon, 2005-09-05 at 22:08 +0300, Petteri Räty wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> I have a couple of old machines I maintain and emerging and unmerging
> kernel sources take a while because there are so many files. Also one
> set of gentoo sources takes about 230MB of disk space. By removing stuff
> not belonging to x86 I was able to succesfully run make with 58MB/230MB
> removed. The stuff I removed:
> arch/* except i386 and x86_64
> include/asm-* expect asm-generic, asm-i386 and asm-x86_64


> So I propose we implement a minimal USE flag in the kernel-2 eclass that
> would make the cleaning I succested before the merging. One problem
> coming from the clean is that make clean does not work and at the moment
> it is run before unmerging, which is of course a good thing. If the
> kernel devs think this is a good idea, I can make an implementation for
> this.

Would something like an unpack.mask/UNPACK_MASK="paths/ wildcards.."
work for you? Perhaps you can simply just take advantage of tar's
--exclude=/-e options in the unpack() function of ebuild.sh when
USERLAND == GNU.  Untested patch attached if you want to play
with/perfect the idea. By excluding it from tar you should be able to
save space/cpu/ram all the way around the board while getting the same
end result.


[-- Attachment #2: unpack_mask.patch --]
[-- Type: text/x-patch, Size: 398 bytes --]

--- /usr/bin/ebuild.sh	2005-09-08 13:48:02.000000000 -0400
+++ ebuild.sh	2005-09-08 13:51:13.000000000 -0400
@@ -346,8 +346,9 @@
 	if [ "$USERLAND" == "BSD" ]; then
 		tarvars=""
 	else
-		tarvars="--no-same-owner"	
-	fi	
+		tarvars="--no-same-owner"
+		[[ "$UNPACK_MASK" != "" ]] && tarvars="${tarvars} -e \'${UNPACK_MASK}\'"
+	fi
 
 	[ -z "$*" ] && die "Nothing passed to the 'unpack' command"
 

Re: [gentoo-dev] USE="minimal" for kernel sources

[Diego 'Flameeyes' Pettenò]

[-- Attachment #1: Type: text/plain, Size: 390 bytes --]

On Thursday 08 September 2005 20:10, solar wrote:
>  Perhaps you can simply just take advantage of tar's
> --exclude=/-e options in the unpack() function of ebuild.sh when
> USERLAND == GNU
tar --exclude/-e is supported by both bsdtar and gtar.

-- 
Diego "Flameeyes" Pettenò
Gentoo Developer - http://dev.gentoo.org/~flameeyes/
(Gentoo/FreeBSD, Video, Gentoo/AMD64, Sound, PAM)

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] USE="minimal" for kernel sources

[John Mylchreest]

[-- Attachment #1: Type: text/plain, Size: 1019 bytes --]

For the record, there is a bug open for this. (#64009)
Personally, I'm not keen on the idea.
the only way which we can do this is by detecting which arch we are
installing the sources, for, which immediately means many installs of
USE=minimal are not the same.

There are plenty of other reasons I can go into, but if anyone feels
strongly to push this change, then feel free to reply with justification
as to why. Technical info to back it up as well please :)

- John

On Thu, 2005-09-08 at 20:17 +0200, Diego 'Flameeyes' Pettenò wrote:
> On Thursday 08 September 2005 20:10, solar wrote:
> >  Perhaps you can simply just take advantage of tar's
> > --exclude=/-e options in the unpack() function of ebuild.sh when
> > USERLAND == GNU
> tar --exclude/-e is supported by both bsdtar and gtar.
> 
-- 
Role:            Gentoo Linux Kernel Lead
Gentoo Linux:    http://www.gentoo.org
Public Key:      gpg --recv-keys 9C745515
Key fingerprint: A0AF F3C8 D699 A05A EC5C  24F7 95AA 241D 9C74 5515


[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] USE="minimal" for kernel sources

[Jan Kundrát]

[-- Attachment #1: Type: text/plain, Size: 495 bytes --]

John Mylchreest wrote:
> For the record, there is a bug open for this. (#64009)
> Personally, I'm not keen on the idea.
> the only way which we can do this is by detecting which arch we are
> installing the sources, for, which immediately means many installs of
> USE=minimal are not the same.

Er, but why is this a problem? Does it matter that the package will
install different files on x86 than on mips? Or am I just overlooking
the point?

-jkt

-- 
cd /local/pub && more beer > /dev/mouth

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 256 bytes --]

Re: [gentoo-dev] USE="minimal" for kernel sources

[John Mylchreest]

[-- Attachment #1: Type: text/plain, Size: 1290 bytes --]

On Thu, 2005-09-08 at 22:14 +0200, Jan Kundrát wrote:

> Er, but why is this a problem? Does it matter that the package will
> install different files on x86 than on mips? Or am I just overlooking
> the point?

In general, there is no obvious technical reason against individual
installs differing from one another, however from a support and QA point
of view it makes it a much less trivial issue. At the end of the day,
when it comes to USE=minimal no-one can fully confirm what does, and
does not exist (can cause breakages may I add) when it comes to
supporting a bug, and also we can't promise it wont destroy an arch tree
which you need. I'm thinking obscure (or not quite so) architecture.
Pegasos, Sun, Sparc, SH, arm, etc.

Although Kbuild is more than capable of functioning with only the
required arch tree, what happens when it comes to things like
cross-compile, not just of the sources but of anything else which might
use them later on? ipw2200? nvidia? alsa-drivers?

Just a bit of a worry to me, and not something I would really like to
endorse.

-- 
Role:            Gentoo Linux Kernel Lead
Gentoo Linux:    http://www.gentoo.org
Public Key:      gpg --recv-keys 9C745515
Key fingerprint: A0AF F3C8 D699 A05A EC5C  24F7 95AA 241D 9C74 5515


[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] USE="minimal" for kernel sources

[Jan Kundrát]

[-- Attachment #1: Type: text/plain, Size: 388 bytes --]

John Mylchreest wrote:
> In general, there is no obvious technical reason against individual
> installs differing from one another, however from a support and QA point
> of view it makes it a much less trivial issue.

Well, I was talking about <10 variants (one for each supported arch),
but iggy's explanation seems reasonable. Thanks.

-jkt

-- 
cd /local/pub && more beer > /dev/mouth

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 256 bytes --]

Re: [gentoo-dev] USE="minimal" for kernel sources

[Brian Jackson]

At one time we had a problem with gentoo sources having way too many use flags 
and patches which lead to there being an incalculable number of ways that gentoo 
-sources could turn out. It was a pita to maintain. A pita to troubleshoot. 
There were weird bugs that we couldn't reproduce easily. Basically, it was one 
of the worst packages to maintain in Gentoo. We had to adopt an attitude of 
complete loathing for kernel source packages having multiple outcomes. This 
loathing is probably still embedded somewhere in all of us that have been on the 
kernel team at one point or another.

--Iggy

Jan Kundrát wrote:
> John Mylchreest wrote:
> 
>>For the record, there is a bug open for this. (#64009)
>>Personally, I'm not keen on the idea.
>>the only way which we can do this is by detecting which arch we are
>>installing the sources, for, which immediately means many installs of
>>USE=minimal are not the same.
> 
> 
> Er, but why is this a problem? Does it matter that the package will
> install different files on x86 than on mips? Or am I just overlooking
> the point?
> 
> -jkt
> 

-- 
I top post... suck it
-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] USE="minimal" for kernel sources

[Tom Fredrik Blenning Klaussen]

On Thu, 2005-09-08 at 20:01 +0100, John Mylchreest wrote:
> For the record, there is a bug open for this. (#64009)
> Personally, I'm not keen on the idea.
> the only way which we can do this is by detecting which arch we are
> installing the sources, for, which immediately means many installs of
> USE=minimal are not the same.

I'm the reporter of the above mentioned bug (which for the record
was /only 1 year old/ September 14. I love the response time :-) )
I was just made aware of this discussion, so sorry about the late
response.

I really can't see the problem with several installs being dissimilar.

> There are plenty of other reasons I can go into, but if anyone feels
> strongly to push this change, then feel free to reply with justification
> as to why. Technical info to back it up as well please :)

The only real difficulty I can see is that the kernel-devs sometimes
pull includes from other arcs. This will cause compile errors, and under
no circumstances any runtime problems. If the use flag also comes with a
warning that use is on your own peril, and support is not given. Too bad
for the people ignoring the warning.
If we find includes from wrong trees, this should be reported upstream,
and we actually gain a valuable tool for reporting some lesser errors in
the kernel.

If for some reason some other package should need this arch specific
stuff, I can not see how this should yield any other result than a
"clean" compile error. If someone could give a counter-example please
enlighten me.

I have my own implementation of this in the kernel-2.eclass. For the
cross-platform issue, it is implemented by requiring each arch to
specify it's needed arch specific directories. If none is specified, the
use flag has no effect.

As for the space saving effect which is the _only_ benefit, this is
quite a significant one if one's system is like mine a 2GiB system, and
a single source takes about 250MiB. The space saving effect is close to
50MiB per source, about 2.5% of the entire system.

> On Thu, 2005-09-08 at 20:17 +0200, Diego 'Flameeyes' Pettenò wrote:
> > On Thursday 08 September 2005 20:10, solar wrote:
> > >  Perhaps you can simply just take advantage of tar's
> > > --exclude=/-e options in the unpack() function of ebuild.sh when
> > > USERLAND == GNU
> > tar --exclude/-e is supported by both bsdtar and gtar.
> > 

-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] USE="minimal" for kernel sources

[Alec Warner]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Tom Fredrik Blenning Klaussen wrote:
> On Thu, 2005-09-08 at 20:01 +0100, John Mylchreest wrote:
> 
>>For the record, there is a bug open for this. (#64009)
>>Personally, I'm not keen on the idea.
>>the only way which we can do this is by detecting which arch we are
>>installing the sources, for, which immediately means many installs of
>>USE=minimal are not the same.
> 
> 
> I'm the reporter of the above mentioned bug (which for the record
> was /only 1 year old/ September 14. I love the response time :-) )
> I was just made aware of this discussion, so sorry about the late
> response.
> 
> I really can't see the problem with several installs being dissimilar.
> 
> 
>>There are plenty of other reasons I can go into, but if anyone feels
>>strongly to push this change, then feel free to reply with justification
>>as to why. Technical info to back it up as well please :)
> 
> 
> The only real difficulty I can see is that the kernel-devs sometimes
> pull includes from other arcs. This will cause compile errors, and under
> no circumstances any runtime problems. If the use flag also comes with a
> warning that use is on your own peril, and support is not given. Too bad
> for the people ignoring the warning.
I could see some sort of pmasked ebuild that did this.  Other than that
though, I wouldn't want users to be able to touch it.  It's well known
that pmasked stuff isn't supported.  One can't add a use flag to an
ebuild and then turn around and say oh that use flag isn't supported.
If it's not supported it shouldn't be there.

Also might want to submit the ebuild to breakmygentoo or some other
overlay.

- -Alec Warner (antarus)

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org
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=xV+j
-----END PGP SIGNATURE-----
-- 
gentoo-dev@gentoo.org mailing list

Two-level USE-flag system VAR: [gentoo-dev] USE="minimal" for kernel sources

[Tom Fredrik Blenning Klaussen]

On Wed, 2005-09-21 at 18:07 -0400, Alec Warner wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Tom Fredrik Blenning Klaussen wrote:
> > On Thu, 2005-09-08 at 20:01 +0100, John Mylchreest wrote:
> > 
> >>For the record, there is a bug open for this. (#64009)
> >>Personally, I'm not keen on the idea.
> >>the only way which we can do this is by detecting which arch we are
> >>installing the sources, for, which immediately means many installs of
> >>USE=minimal are not the same.
> > 
> > 
> > I'm the reporter of the above mentioned bug (which for the record
> > was /only 1 year old/ September 14. I love the response time :-) )
> > I was just made aware of this discussion, so sorry about the late
> > response.
> > 
> > I really can't see the problem with several installs being dissimilar.
> > 
> > 
> >>There are plenty of other reasons I can go into, but if anyone feels
> >>strongly to push this change, then feel free to reply with justification
> >>as to why. Technical info to back it up as well please :)
> > 
> > 
> > The only real difficulty I can see is that the kernel-devs sometimes
> > pull includes from other arcs. This will cause compile errors, and under
> > no circumstances any runtime problems. If the use flag also comes with a
> > warning that use is on your own peril, and support is not given. Too bad
> > for the people ignoring the warning.
> I could see some sort of pmasked ebuild that did this.  Other than that
> though, I wouldn't want users to be able to touch it.  It's well known
> that pmasked stuff isn't supported.  One can't add a use flag to an
> ebuild and then turn around and say oh that use flag isn't supported.
> If it's not supported it shouldn't be there.

Actually  a method for masking USE flags could be a good idea. Now, as
said earlier the natural way to do this is in the kernel-2.eclass. So I
don't see how one could pmask that.

Why are you afraid of users coming near this. I don't expect this to
break anything at all, but even if the errors generated are straight
forward. The average gentoo users are not stupid. In fact if you have
come so far as to adjust something beyond the most basic USE flags at
all, you're probably advanced enough to deciphre such a message. (It
would be nice to have some knowledge of who the average gentoo user is
though.)

Now as for the USE flag system. It has actually become so big that it's
difficult to use it effectively. I would actually suggest that a two
level system of USE flags could be employed. Something like
wtk/gtk (Windowing Toolkit / gtk)
wtk/kde (Windowing Toolkit / kde)

There could also be another category
experimental/minimal

Using a flag from experimental would add a warning to the pacakge that
uses it, and an implicit understanding that tweaking of the experimental
use flags is strongly discouraged.

> Also might want to submit the ebuild to breakmygentoo or some other
> overlay.

I'll consider it, but as mentioned above it's really a change to an
eclass.

Sincerely
-- 
Tom Fredrik Klaussen
-- 
gentoo-dev@gentoo.org mailing list

Re: Two-level USE-flag system VAR: [gentoo-dev] USE="minimal" for kernel sources

[warnera6]

Tom Fredrik Blenning Klaussen wrote:

 > The average gentoo users are not stupid.
Many people would not agree with that statement ;)

> come so far as to adjust something beyond the most basic USE flags at
> all, you're probably advanced enough to deciphre such a message. (It
> would be nice to have some knowledge of who the average gentoo user is
> though.)
> 
> Now as for the USE flag system. It has actually become so big that it's
> difficult to use it effectively.
   There is a USE flag group GLEP that is being implemented...sometime ;)
   I don't think masking USE flags has come up...*pokes portage people*
>>Also might want to submit the ebuild to breakmygentoo or some other
>>overlay.
> 
> 
> I'll consider it, but as mentioned above it's really a change to an
> eclass.
You can put eclasses in the overlay as well IIRC.

-Alec Warner (antarus)
-- 
gentoo-dev@gentoo.org mailing list

Re: Two-level USE-flag system VAR: [gentoo-dev] USE="minimal" for kernel sources

[Jason Stubbs]

[-- Attachment #1: Type: text/plain, Size: 699 bytes --]

On Friday 23 September 2005 05:28, Tom Fredrik Blenning Klaussen wrote:
> Now as for the USE flag system. It has actually become so big that it's
> difficult to use it effectively. I would actually suggest that a two
> level system of USE flags could be employed. Something like
> wtk/gtk (Windowing Toolkit / gtk)
> wtk/kde (Windowing Toolkit / kde)

This is just arbitrary grouping as far as USE flags themselves go. Rather 
than changing the name of the flags, why not just split the flags that are 
in use.desc into categories separated by comments?

# some category
use ...
use ...
...

# Windowing Toolkits
gtk ...
kde ...

# some other category
...

-- 
Jason Stubbs

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

[gentoo-dev] Re: Two-level USE-flag system VAR: USE="minimal" for kernel sources

[Duncan]

Jason Stubbs posted <200509231019.16895.jstubbs@gentoo.org>, excerpted
below,  on Fri, 23 Sep 2005 10:19:14 +0900:

> On Friday 23 September 2005 05:28, Tom Fredrik Blenning Klaussen wrote:
>> Now as for the USE flag system. It has actually become so big that it's
>> difficult to use it effectively. I would actually suggest that a two
>> level system of USE flags could be employed. Something like
>> wtk/gtk (Windowing Toolkit / gtk)
>> wtk/kde (Windowing Toolkit / kde)
> 
> This is just arbitrary grouping as far as USE flags themselves go. Rather 
> than changing the name of the flags, why not just split the flags that are 
> in use.desc into categories separated by comments?
> 
> # some category
> use ...
> use ...
> ...
> 
> # Windowing Toolkits
> gtk ...
> kde ...
> 
> # some other category
> ...

The problem as I see it with comment-categories for USE flags is that it
doesn't well match how USE flags (and looking up USE flag descriptions)
are actually used.

TFBKlaussen's proposal would make it immediately obvious from an emerge
--verbose --ask (or --pretend) what category was involved.  Commenting
use.desc (and use.local.desc) doesn't have that advantage.

Additionally, when I look up a description, it's usually by grepping
use.(local.)desc, and I suppose many others work similarly.  I/we don't
care about all the /other descriptions, only the one we are wondering
about.  Putting additional information in a comment line ?? lines above
the flag and description in question would /not/ be helpful.  OTOH, using
a category/flag arrangement would be somewhat of a description of its own,
meaning the description could be shortened, and the line would be no
longer than it is currently.  (With 80-char screen widths, this can be an
issue.)

OTOH, it's obviously yet /another/ thing for portage devs to work on, and
portage is /supposed/ to be in feature request freeze ATM...  I like the
idea, but whether the benefits of putting it on the current feature list
outweigh the costs of putting it off, is something I'm not going to even
pretend I want to evaluate.  =8^|  If you portage devs believe it's easy
to "make it so", perhaps further discussion is warranted.  If not, I'm
not in favor of putting off the next portage yet /again/ to make it
happen, tho it'd certainly be nice to have, so I'd say it's not even worth
further discussion ATM.  JMHO...

-- 
Duncan - List replies preferred.   No HTML msgs.
"Every nonfree program has a lord, a master --
and if you use the program, he is your master."  Richard Stallman in
http://www.linuxdevcenter.com/pub/a/linux/2004/12/22/rms_interview.html


-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Re: Two-level USE-flag system VAR: USE="minimal" for kernel sources

[Jason Stubbs]

[-- Attachment #1: Type: text/plain, Size: 1080 bytes --]

On Friday 23 September 2005 15:36, Duncan wrote:
> OTOH, it's obviously yet /another/ thing for portage devs to work on, and
> portage is /supposed/ to be in feature request freeze ATM...  I like the
> idea, but whether the benefits of putting it on the current feature list
> outweigh the costs of putting it off, is something I'm not going to even
> pretend I want to evaluate.  =8^|  If you portage devs believe it's easy
> to "make it so", perhaps further discussion is warranted.  If not, I'm
> not in favor of putting off the next portage yet /again/ to make it
> happen, tho it'd certainly be nice to have, so I'd say it's not even
> worth further discussion ATM.  JMHO...

There's absolutely no work required on portage to support USE flag 
"categories". There's nothing preventing a "/" character from appearing in 
a USE flag, hence the support is there already. The work would come in 
adjusting every ebuild to use the different name as well as killing 
backward compatibility with already installed packages and binary packages. 

-- 
Jason Stubbs

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

[gentoo-dev] Re: Re: Two-level USE-flag system VAR: USE="minimal" for kernel sources

[Duncan]

Jason Stubbs posted <200509231644.47078.jstubbs@gentoo.org>, excerpted
below,  on Fri, 23 Sep 2005 16:44:44 +0900:

> There's absolutely no work required on portage to support USE flag
> "categories". There's nothing preventing a "/" character from appearing in
> a USE flag, hence the support is there already.

It's not treated as some sort of metacharacter or otherwise currently
input filtered, then, I take it.

> The work would come in adjusting every ebuild to use the different
> name as well as killing backward compatibility with already installed
> packages and binary packages.

... and existing make.confs and package.uses, of course...  Good point.

It'd certainly be lots of work, and I can envision the discussions over
whether various flags belong in this or that category, and where to put
the ones that don't seem to fit a category, but it might be worth it in
terms of clarity and ease of use (or maybe not, as the PHP example points
out).

-- 
Duncan - List replies preferred.   No HTML msgs.
"Every nonfree program has a lord, a master --
and if you use the program, he is your master."  Richard Stallman in
http://www.linuxdevcenter.com/pub/a/linux/2004/12/22/rms_interview.html


-- 
gentoo-dev@gentoo.org mailing list

Re: Two-level USE-flag system VAR: [gentoo-dev] USE="minimal" for kernel sources

[Thomas de Grenier de Latour]

On Thu, 22 Sep 2005 22:28:35 +0200
Tom Fredrik Blenning Klaussen <bfg-gentoo@blenning.no> wrote:

> Now as for the USE flag system. It has actually become so big
> that it's difficult to use it effectively. I would actually
> suggest that a two level system of USE flags could be employed.
> Something like
> wtk/gtk (Windowing Toolkit / gtk)
> wtk/kde (Windowing Toolkit / kde)

Sounds good on this example, but i'm not sure it would be that easy
and meaningfull on the whole use.desc. I think you would end up
with lot of "this flag should be in that category and not in this
one" discussions (what already happen with packages).  Also, i
would be curious to see the output of an "emerge -pv" on some
highly configurable packages (dev-lang/php comes to mind for
instance), and whether it really improves readibility.

> There could also be another category
> experimental/minimal

If the idea is just to give a "don't use this flag but if you know
what you are doing" warning, then the best imho is simply to
use.mask it in base profile.  And people who know what they are
doing can unmask it from their /etc/portage/profile/use.mask
(syntax is "-flag"). Sure, it should not be named "minimal" in that
case, since "minimal" is not something you want to mask, but rather
"mini-kernel-src" or anything else that sounds like a specific flag.

--
TGL.
-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] USE="minimal" for kernel sources

[twofourtysix]

On 05/09/05, Petteri Räty <betelgeuse@gentoo.org> wrote:
> I have a couple of old machines I maintain and emerging and unmerging
> kernel sources take a while because there are so many files. Also one
> set of gentoo sources takes about 230MB of disk space. By removing stuff
> not belonging to x86 I was able to succesfully run make with 58MB/230MB
> removed. The stuff I removed:
> arch/* except i386 and x86_64
> include/asm-* expect asm-generic, asm-i386 and asm-x86_64

Is this safe? Various parts of the kernel have been known to pull in
things from outside the obvious places. I don't think the source tree
is intended to be used like this, so your proposal could lead to build
errors. Besides, does this really matter? Disk space grows on trees,
and with modern file systems it's not like limited inode counts are an
issue.

-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] USE="minimal" for kernel sources

[Greg KH]

On Thu, Sep 08, 2005 at 10:49:04PM +0100, twofourtysix wrote:
> On 05/09/05, Petteri R?ty <betelgeuse@gentoo.org> wrote:
> > I have a couple of old machines I maintain and emerging and unmerging
> > kernel sources take a while because there are so many files. Also one
> > set of gentoo sources takes about 230MB of disk space. By removing stuff
> > not belonging to x86 I was able to succesfully run make with 58MB/230MB
> > removed. The stuff I removed:
> > arch/* except i386 and x86_64
> > include/asm-* expect asm-generic, asm-i386 and asm-x86_64
> 
> Is this safe?

No it isn't.  Please don't try to do this, it's not worth it.  If disk
space is limited, just build on one box, and install the kernel to the
other one.

Or, put the kernel source on a cd, and build off of it (putting the
objects on your local disk.)  This lets you only use the local disk for
your built objects.

thanks,

greg k-h
-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] USE="minimal" for kernel sources

[warnera6]

Greg KH wrote:
> On Thu, Sep 08, 2005 at 10:49:04PM +0100, twofourtysix wrote:
> 
>>On 05/09/05, Petteri R?ty <betelgeuse@gentoo.org> wrote:
>>
>>>I have a couple of old machines I maintain and emerging and unmerging
>>>kernel sources take a while because there are so many files. Also one
>>>set of gentoo sources takes about 230MB of disk space. By removing stuff
>>>not belonging to x86 I was able to succesfully run make with 58MB/230MB
>>>removed. The stuff I removed:
>>>arch/* except i386 and x86_64
>>>include/asm-* expect asm-generic, asm-i386 and asm-x86_64
>>
>>Is this safe?
> 
> 
> No it isn't.  Please don't try to do this, it's not worth it.  If disk
> space is limited, just build on one box, and install the kernel to the
> other one.
IMHO it is, but not as a USE flag (it will never be stable enough 
without upstream support) but I think many would find the functionality 
useful in a script.  I know I would.  If it works most of the time and 
saves space, there is no reason not trim things.  If it breaks, you 
immediately revert to a normal build.
> 
> Or, put the kernel source on a cd, and build off of it (putting the
> objects on your local disk.)  This lets you only use the local disk for
> your built objects.
> 
> thanks,
> 
> greg k-h

-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] USE="minimal" for kernel sources

[Petteri Räty]

[-- Attachment #1: Type: text/plain, Size: 843 bytes --]

warnera6 wrote:
> 
> IMHO it is, but not as a USE flag (it will never be stable enough
> without upstream support) but I think many would find the functionality
> useful in a script.  I know I would.  If it works most of the time and
> saves space, there is no reason not trim things.  If it breaks, you
> immediately revert to a normal build.
> 

Well this would not give the advantage of cutting down emerge times. Why
the script when you can just turn off the use flag?

>>
>> Or, put the kernel source on a cd, and build off of it (putting the
>> objects on your local disk.)  This lets you only use the local disk for
>> your built objects.

I can always maintain this in my overlay if I want to so that is no
problem. I just thought this might be useful for other people too.

Regards,
Petteri Räty (Betelgeuse)


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 256 bytes --]

Re: [gentoo-dev] USE="minimal" for kernel sources

[Alec Warner]

Petteri Räty wrote:

>warnera6 wrote:
>  
>
>>IMHO it is, but not as a USE flag (it will never be stable enough
>>without upstream support) but I think many would find the functionality
>>useful in a script.  I know I would.  If it works most of the time and
>>saves space, there is no reason not trim things.  If it breaks, you
>>immediately revert to a normal build.
>>
>>    
>>
>
>Well this would not give the advantage of cutting down emerge times. Why
>the script when you can just turn off the use flag?
>
>  
>
Because the USE flag itself doesn't produce stable output and shouldn't 
be in a stable tree, IMHO.  You can't mask use flags ~arch.

>>>Or, put the kernel source on a cd, and build off of it (putting the
>>>objects on your local disk.)  This lets you only use the local disk for
>>>your built objects.
>>>      
>>>
>
>I can always maintain this in my overlay if I want to so that is no
>problem. I just thought this might be useful for other people too.
>
>Regards,
>Petteri Räty (Betelgeuse)
>
>  
>

-- 
gentoo-dev@gentoo.org mailing list