[gentoo-dev] Identifying inherit-only / usable profiles

[gentoo-dev] Identifying inherit-only / usable profiles

[Ciaran McCreesh]

[-- Attachment #1: Type: text/plain, Size: 545 bytes --]

AFAICS there's no way to tell whether a profile is "inherit only" or
whether it's complete enough to be used by an end user. Would it make
sense to have an empty text file called "inherit-only" (for example)
within profiles which aren't usable?

Here's a pretty picture of all the profiles as of whenever I last
synced:

http://dev.gentoo.org/~ciaranm/docs/profile-tree/profile-tree.png

-- 
Ciaran McCreesh : Gentoo Developer (Vim, Fluxbox, Sparc, Mips)
Mail            : ciaranm at gentoo.org
Web             : http://dev.gentoo.org/~ciaranm


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

Re: [gentoo-dev] Identifying inherit-only / usable profiles

[Chris Gianelloni]

[-- Attachment #1: Type: text/plain, Size: 774 bytes --]

On Thu, 2004-12-02 at 19:31 +0000, Ciaran McCreesh wrote:
> AFAICS there's no way to tell whether a profile is "inherit only" or
> whether it's complete enough to be used by an end user. Would it make
> sense to have an empty text file called "inherit-only" (for example)
> within profiles which aren't usable?

I was always under the impression that all profiles should be usable.
If that is not the case, then I 100% agree that there should be some way
of marking a profile as unusable.

> Here's a pretty picture of all the profiles as of whenever I last
> synced:
> 
> http://dev.gentoo.org/~ciaranm/docs/profile-tree/profile-tree.png

Ohh... pretty

-- 
Chris Gianelloni
Release Engineering - Operational/QA Manager
Games - Developer
Gentoo Linux

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

Re: [gentoo-dev] Identifying inherit-only / usable profiles

[Ciaran McCreesh]

[-- Attachment #1: Type: text/plain, Size: 858 bytes --]

On Thu, 02 Dec 2004 16:12:40 -0500 Chris Gianelloni
<wolf31o2@gentoo.org> wrote:
| On Thu, 2004-12-02 at 19:31 +0000, Ciaran McCreesh wrote:
| > AFAICS there's no way to tell whether a profile is "inherit only" or
| > whether it's complete enough to be used by an end user. Would it
| > make sense to have an empty text file called "inherit-only" (for
| > example) within profiles which aren't usable?
| 
| I was always under the impression that all profiles should be usable.
| If that is not the case, then I 100% agree that there should be some
| way of marking a profile as unusable.

Well, you couldn't really sanely run a system with your profile symlink
pointing to 'base' or 'default-linux'.

-- 
Ciaran McCreesh : Gentoo Developer (Vim, Fluxbox, Sparc, Mips)
Mail            : ciaranm at gentoo.org
Web             : http://dev.gentoo.org/~ciaranm


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

Re: [gentoo-dev] Identifying inherit-only / usable profiles

[Chris Gianelloni]

[-- Attachment #1: Type: text/plain, Size: 1592 bytes --]

On Thu, 2004-12-02 at 21:25 +0000, Ciaran McCreesh wrote:
> On Thu, 02 Dec 2004 16:12:40 -0500 Chris Gianelloni
> <wolf31o2@gentoo.org> wrote:
> | On Thu, 2004-12-02 at 19:31 +0000, Ciaran McCreesh wrote:
> | > AFAICS there's no way to tell whether a profile is "inherit only" or
> | > whether it's complete enough to be used by an end user. Would it
> | > make sense to have an empty text file called "inherit-only" (for
> | > example) within profiles which aren't usable?
> | 
> | I was always under the impression that all profiles should be usable.
> | If that is not the case, then I 100% agree that there should be some
> | way of marking a profile as unusable.
> 
> Well, you couldn't really sanely run a system with your profile symlink
> pointing to 'base' or 'default-linux'.

...and like every time you assume something, somebody has to come out
and point out the fact that you're an ass.

Yeah, I hadn't considered default-linux or base.  Though, to be honest,
I bet you would have a working system with either of them, or at leats
you should, even if it did keep from having arch or release-specific
changes.  After all, most of it would be the changes in virtuals, which
so long as you emerged the right programs to satisfy the virtuals for
your arch, you should be fine with either of these.  However, you are
definitely correct that you can no select one of these profiles, do an
"emerge system" and expect it to work without some intervention.

-- 
Chris Gianelloni
Release Engineering - Operational/QA Manager
Games - Developer
Gentoo Linux

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

Re: [gentoo-dev] Identifying inherit-only / usable profiles

[Jason Stubbs]

On Friday 03 December 2004 22:14, Chris Gianelloni wrote:
> On Thu, 2004-12-02 at 21:25 +0000, Ciaran McCreesh wrote:
> > On Thu, 02 Dec 2004 16:12:40 -0500 Chris Gianelloni
> >
> > <wolf31o2@gentoo.org> wrote:
> > | On Thu, 2004-12-02 at 19:31 +0000, Ciaran McCreesh wrote:
> > | > AFAICS there's no way to tell whether a profile is "inherit only" or
> > | > whether it's complete enough to be used by an end user. Would it
> > | > make sense to have an empty text file called "inherit-only" (for
> > | > example) within profiles which aren't usable?
> > |
> > | I was always under the impression that all profiles should be usable.
> > | If that is not the case, then I 100% agree that there should be some
> > | way of marking a profile as unusable.
> >
> > Well, you couldn't really sanely run a system with your profile symlink
> > pointing to 'base' or 'default-linux'.
>
> ...and like every time you assume something, somebody has to come out
> and point out the fact that you're an ass.
>
> Yeah, I hadn't considered default-linux or base.  Though, to be honest,
> I bet you would have a working system with either of them, or at leats
> you should, even if it did keep from having arch or release-specific
> changes.  After all, most of it would be the changes in virtuals, which
> so long as you emerged the right programs to satisfy the virtuals for
> your arch, you should be fine with either of these.  However, you are
> definitely correct that you can no select one of these profiles, do an
> "emerge system" and expect it to work without some intervention.

# ln -sf /usr/portage/profiles/base /etc/make.profile
# emerge -p system

These are the packages that I would merge, in order:

!!! ARCH is not set... Are you missing the /etc/make.profile symlink?
!!! Is the symlink correct? Is your portage tree complete?


Regards,
Jason Stubbs

--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Identifying inherit-only / usable profiles

[Ciaran McCreesh]

[-- Attachment #1: Type: text/plain, Size: 819 bytes --]

On Fri, 3 Dec 2004 22:26:17 +0900 Jason Stubbs <jstubbs@gentoo.org>
wrote:
| # ln -sf /usr/portage/profiles/base /etc/make.profile
| # emerge -p system
| 
| These are the packages that I would merge, in order:
| 
| !!! ARCH is not set... Are you missing the /etc/make.profile symlink?
| !!! Is the symlink correct? Is your portage tree complete?

Indeed. So... Can anyone think of a better name than "inherit-only"? Can
anyone think of a good way of indicating this other than by an empty
file? Would it be reasonable for me to file a bug asking that at some
point in the future portage detects the existence of said file and
displays a better message instead?

-- 
Ciaran McCreesh : Gentoo Developer (Vim, Fluxbox, Sparc, Mips)
Mail            : ciaranm at gentoo.org
Web             : http://dev.gentoo.org/~ciaranm


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

Re: [gentoo-dev] Identifying inherit-only / usable profiles

[Jason Stubbs]

On Friday 03 December 2004 23:13, Ciaran McCreesh wrote:
> On Fri, 3 Dec 2004 22:26:17 +0900 Jason Stubbs <jstubbs@gentoo.org>
>
> wrote:
> | # ln -sf /usr/portage/profiles/base /etc/make.profile
> | # emerge -p system
> |
> | These are the packages that I would merge, in order:
> |
> | !!! ARCH is not set... Are you missing the /etc/make.profile symlink?
> | !!! Is the symlink correct? Is your portage tree complete?
>
> Indeed. So... Can anyone think of a better name than "inherit-only"? Can
> anyone think of a good way of indicating this other than by an empty
> file? Would it be reasonable for me to file a bug asking that at some
> point in the future portage detects the existence of said file and
> displays a better message instead?

Heh.. It'd take about 2 minutes to write and test the 3 lines of code to 
implement said enhancement. The question is, is this the correct enhancement. 
If so, why? I'd be more inclined to use a file to define a profile as being 
"valid" rather than "invalid".

Perhaps a "description" file that describes the purpose of that profile. 
Perhaps even a metadata.xml that contains the purpose of that specific 
profile directory within its cascade as well as a flag indicating to portage 
whether it's intented to be linked to.

Anyway, whatever happens - please, no bugs!! ;)

Regards,
Jason Stubbs

--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Identifying inherit-only / usable profiles

[Ciaran McCreesh]

[-- Attachment #1: Type: text/plain, Size: 1314 bytes --]

On Fri, 3 Dec 2004 23:32:14 +0900 Jason Stubbs <jstubbs@gentoo.org>
wrote:
| > Indeed. So... Can anyone think of a better name than "inherit-only"?
| > Can anyone think of a good way of indicating this other than by an
| > empty file? Would it be reasonable for me to file a bug asking that
| > at some point in the future portage detects the existence of said
| > file and displays a better message instead?
| 
| Heh.. It'd take about 2 minutes to write and test the 3 lines of code
| to implement said enhancement. The question is, is this the correct
| enhancement. If so, why? I'd be more inclined to use a file to define
| a profile as being "valid" rather than "invalid".

Sure, that'd work too, I just couldn't think of a suitably clear name
for the file. Erm, I mean, I was thinking of backwards compatibility.

| Perhaps a "description" file that describes the purpose of that
| profile. Perhaps even a metadata.xml that contains the purpose of that
| specific profile directory within its cascade as well as a flag
| indicating to portage whether it's intented to be linked to.

That'd be nice, apart from the xml bit (pita to parse in bash...).

-- 
Ciaran McCreesh : Gentoo Developer (Vim, Fluxbox, Sparc, Mips)
Mail            : ciaranm at gentoo.org
Web             : http://dev.gentoo.org/~ciaranm


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

Re: [gentoo-dev] Identifying inherit-only / usable profiles

[Chris Gianelloni]

[-- Attachment #1: Type: text/plain, Size: 915 bytes --]

On Fri, 2004-12-03 at 23:32 +0900, Jason Stubbs wrote:
> Heh.. It'd take about 2 minutes to write and test the 3 lines of code to 
> implement said enhancement. The question is, is this the correct enhancement. 
> If so, why? I'd be more inclined to use a file to define a profile as being 
> "valid" rather than "invalid".
> 
> Perhaps a "description" file that describes the purpose of that profile. 
> Perhaps even a metadata.xml that contains the purpose of that specific 
> profile directory within its cascade as well as a flag indicating to portage 
> whether it's intented to be linked to.

I tend to like something like this better.  Then we could use the same
file to notify of profile deprecation, whether a profile is
inherit-only, or anything else that we come up with in the future.

-- 
Chris Gianelloni
Release Engineering - Operational/QA Manager
Games - Developer
Gentoo Linux

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

Re: [gentoo-dev] Identifying inherit-only / usable profiles

[Donnie Berkholz]

[-- Attachment #1: Type: text/plain, Size: 511 bytes --]

On Fri, 2004-12-03 at 23:32 +0900, Jason Stubbs wrote:
> Heh.. It'd take about 2 minutes to write and test the 3 lines of code to 
> implement said enhancement. The question is, is this the correct enhancement. 
> If so, why? I'd be more inclined to use a file to define a profile as being 
> "valid" rather than "invalid".

That's kinda like defining a profile as "non-deprecated" rather than
"deprecated." To me, it doesn't make sense to say when something works
properly -- that should be expected.

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

Re: [gentoo-dev] Identifying inherit-only / usable profiles

[Luke-Jr]

[-- Attachment #1: Type: text/plain, Size: 994 bytes --]

On Friday 03 December 2004 5:19 pm, Donnie Berkholz wrote:
> On Fri, 2004-12-03 at 23:32 +0900, Jason Stubbs wrote:
> > Heh.. It'd take about 2 minutes to write and test the 3 lines of code to
> > implement said enhancement. The question is, is this the correct
> > enhancement. If so, why? I'd be more inclined to use a file to define a
> > profile as being "valid" rather than "invalid".
>
> That's kinda like defining a profile as "non-deprecated" rather than
> "deprecated." To me, it doesn't make sense to say when something works
> properly -- that should be expected.

A better comparison would be like defining them as "completed" instead of 
"incomplete".
If a valid-marker is used, it can (and probably should) be inherited. If an 
invalid-marker is used, you almost *never* want to inherit it.
Alternatively, it would be possible to mark profiles as either valid or 
invalid and inherit the value from parents.
-- 
Luke-Jr
Developer, Utopios
http://utopios.org/

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

Re: [gentoo-dev] Identifying inherit-only / usable profiles

[Ciaran McCreesh]

[-- Attachment #1: Type: text/plain, Size: 1038 bytes --]

On Fri, 3 Dec 2004 17:48:51 +0000 Luke-Jr <luke-jr@utopios.org> wrote:
| A better comparison would be like defining them as "completed" instead
| of "incomplete".
| If a valid-marker is used, it can (and probably should) be inherited.

No no no. A profile might be valid have some subprofiles which are not
themselves directly valid. Random example, we might have something
vaguely like (example only, not suggested as anything like an actual
possible implementation):

default-linux/mips/2005.0/                       valid
default-linux/mips/2005.0/multilib               not valid
default-linux/mips/2005.0/multilib/n32-o32/      valid
default-linux/mips/2005.0/multilib/n64-n32-o32/  valid

Ok, that's a pretty bad example. In fact it's a terrible example, but I
can't think of a better one off the top of my head. Chances are sooner
or later it'll end up happening though...

-- 
Ciaran McCreesh : Gentoo Developer (Vim, Fluxbox, Sparc, Mips)
Mail            : ciaranm at gentoo.org
Web             : http://dev.gentoo.org/~ciaranm


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

Re: [gentoo-dev] Identifying inherit-only / usable profiles

[Luke-Jr]

On Friday 03 December 2004 6:00 pm, Ciaran McCreesh wrote:
> On Fri, 3 Dec 2004 17:48:51 +0000 Luke-Jr <luke-jr@utopios.org> wrote:
> | A better comparison would be like defining them as "completed" instead
> | of "incomplete".
> | If a valid-marker is used, it can (and probably should) be inherited.
>
> No no no. A profile might be valid have some subprofiles which are not
> themselves directly valid. Random example, we might have something
> vaguely like (example only, not suggested as anything like an actual
> possible implementation):
>
> <snip>
>
> Ok, that's a pretty bad example. In fact it's a terrible example, but I
> can't think of a better one off the top of my head. Chances are sooner
> or later it'll end up happening though...

Which is why it might be better to have profiles be either valid-marked or 
invalid-marked and default to inheriting.
-- 
Luke-Jr
Developer, Utopios
http://utopios.org/

--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Identifying inherit-only / usable profiles

[Ed Grimm]

On Fri, 3 Dec 2004, Luke-Jr wrote:

> On Friday 03 December 2004 5:19 pm, Donnie Berkholz wrote:
>> On Fri, 2004-12-03 at 23:32 +0900, Jason Stubbs wrote:
>>> Heh.. It'd take about 2 minutes to write and test the 3 lines of code to
>>> implement said enhancement. The question is, is this the correct
>>> enhancement. If so, why? I'd be more inclined to use a file to define a
>>> profile as being "valid" rather than "invalid".
>>
>> That's kinda like defining a profile as "non-deprecated" rather than
>> "deprecated." To me, it doesn't make sense to say when something works
>> properly -- that should be expected.
>
> A better comparison would be like defining them as "completed" instead of
> "incomplete".
> If a valid-marker is used, it can (and probably should) be inherited. If an
> invalid-marker is used, you almost *never* want to inherit it.
> Alternatively, it would be possible to mark profiles as either valid or
> invalid and inherit the value from parents.

We want to inherit valid/invalid for what reason?

I would think that each profile would specify on its own whether or not
it was valid; I'd hope the preference would be to specify invalid, and
assume valid, simply because the vast majority of profiles should be
valid.  It seems to me that maintaining the valid list should be more
work than maintaining the invalid list, and if the reverse is true,
there is a problem in the way we are maintaining profiles.

Semantics can be argued either way; I personally feel the unix return
code is the best metaphor - there's one success code, but many failure
codes.  A profile can be supported, unsupported, deprecated, incomplete,
or broken.  There's quite possibly a number of other states as well.

Ed

--
gentoo-dev@gentoo.org mailing list

[gentoo-dev] Re: Identifying inherit-only / usable profiles

[Duncan]

Ed Grimm posted <Pine.LNX.4.60.0412052148100.5557@mbeq.rq.iarg>, excerpted
below,  on Sun, 05 Dec 2004 21:59:17 +0000:

> It seems to me that maintaining the valid list should be more
> work than maintaining the invalid list, and if the reverse is true,
> there is a problem in the way we are maintaining profiles.
> 
> Semantics can be argued either way; I personally feel the unix return
> code is the best metaphor - there's one success code, but many failure
> codes.  A profile can be supported, unsupported, deprecated, incomplete,
> or broken.  There's quite possibly a number of other states as well.

FWIW, as a user...

I like your error code metaphor, but would favor a single file with
status, valid or form of invalid, with a specific error code if the file
isn't there at all.

The reasoning is this.  One can never tell where a user is going to point
the symlink.  What if it points to /usr, for instance.  

Having a single status file allows portage a far more intelligent error
than the current 'Is there a symlink at all and is it pointed to a valid
location?' error.  We can then have multiple error codes fitting the
situation:

0/valid

These would be various not a valid symlink errors

101/no /etc/make.profile symlink
102/file or dir, not symlink
103/symlink points to non-existent file/dir
104/symlink points to file, not dir
105/incomplete, no status file, is this a portage profile dir?

201/Status file present but profile incomplete due to lack of other info

And the following warnings, overridable with an appropriate switch, for
testing, or temp use of a a depreciated profile, for instance

501/masked as known-broken, unsupported
502/masked for testing, unsupported
503/masked as depreciated, unsupported, message points to recommended
upgrade

Of course, the error code numbers would be adjusted to fit portages
current error code structure.

-- 
Duncan - List replies preferred.   No HTML msgs.
"They that can give up essential liberty to obtain a little
temporary safety, deserve neither liberty nor safety." --
Benjamin Franklin



--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Identifying inherit-only / usable profiles

[Paul de Vrieze]

[-- Attachment #1: Type: text/plain, Size: 565 bytes --]

On Friday 03 December 2004 15:13, Ciaran McCreesh wrote:
> Indeed. So... Can anyone think of a better name than "inherit-only"?
> Can anyone think of a good way of indicating this other than by an
> empty file? Would it be reasonable for me to file a bug asking that at
> some point in the future portage detects the existence of said file and
> displays a better message instead?

What about the file that is used to indicate deprecated profiles?

Paul

-- 
Paul de Vrieze
Gentoo Developer
Mail: pauldv@gentoo.org
Homepage: http://www.devrieze.net

[-- Attachment #2: Type: application/pgp-signature, Size: 190 bytes --]

Re: [gentoo-dev] Identifying inherit-only / usable profiles

[Stephen Bennett]

On Fri, 2004-12-03 at 08:14 -0500, Chris Gianelloni wrote:
> Though, to be honest,
> I bet you would have a working system with either of them, or at leats
> you should, even if it did keep from having arch or release-specific
> changes. 

Neither of them set $ARCH, therefore portage refuses to work.


--
gentoo-dev@gentoo.org mailing list