[gentoo-dev] [PATCH] skel.ebuild: Update comments for inherit, SLOT, KEYWORDS.

[gentoo-dev] [PATCH] skel.ebuild: Update comments for inherit, SLOT, KEYWORDS.

[Ulrich Müller]

[-- Attachment #1: Type: text/plain, Size: 1815 bytes --]

epatch() is provided by epatch.eclass now. Also comment the inherit
line, since not every ebuild will use it.

Empty SLOT doesn't disable slots, but is outright illegal in all EAPIs.
Similar for KEYWORDS="*", which isn't "deprecated" but invalid.
---
 skel.ebuild | 9 ++++-----
 1 file changed, 4 insertions(+), 5 deletions(-)

diff --git a/skel.ebuild b/skel.ebuild
index 7ac9dfb7d6d8..4c19b1de4cb8 100644
--- a/skel.ebuild
+++ b/skel.ebuild
@@ -13,9 +13,9 @@
 EAPI=6
 
 # inherit lists eclasses to inherit functions from. For example, an ebuild
-# that needs the epatch function from eutils.eclass won't work without the
+# that needs the epatch function from epatch.eclass won't work without the
 # following line:
-inherit eutils
+#inherit epatch
 #
 # eclasses tend to list descriptions of how to use their functions properly.
 # take a look at /usr/portage/eclass/ for more examples.
@@ -46,7 +46,7 @@ LICENSE=""
 # of each SLOT and remove everything else.
 # Note that normal applications should use SLOT="0" if possible, since
 # there should only be exactly one version installed at a time.
-# DO NOT USE SLOT=""! This tells Portage to disable SLOTs for this package.
+# Do not use SLOT="", because the SLOT variable must not be empty.
 SLOT="0"
 
 # Using KEYWORDS, we can record masking information *inside* an ebuild
@@ -63,8 +63,7 @@ SLOT="0"
 # For binary packages, use -* and then list the archs the bin package
 # exists for.  If the package was for an x86 binary package, then
 # KEYWORDS would be set like this: KEYWORDS="-* x86"
-# DO NOT USE KEYWORDS="*".  This is deprecated and only for backward
-# compatibility reasons.
+# Do not use KEYWORDS="*".  This is not valid in an ebuild context.
 KEYWORDS="~x86"
 
 # Comprehensive list of any and all USE flags leveraged in the ebuild,
-- 
2.15.1

[-- Attachment #2: Type: application/pgp-signature, Size: 490 bytes --]

Re: [gentoo-dev] [PATCH] skel.ebuild: Update comments for inherit, SLOT, KEYWORDS.

[Mike Gilbert]

On Sun, Dec 31, 2017 at 8:31 AM, Ulrich Müller <ulm@gentoo.org> wrote:
> @@ -63,8 +63,7 @@ SLOT="0"
>  # For binary packages, use -* and then list the archs the bin package
>  # exists for.  If the package was for an x86 binary package, then
>  # KEYWORDS would be set like this: KEYWORDS="-* x86"
> -# DO NOT USE KEYWORDS="*".  This is deprecated and only for backward
> -# compatibility reasons.
> +# Do not use KEYWORDS="*".  This is not valid in an ebuild context.
>  KEYWORDS="~x86"

Portage actually does support KEYWORDS="*". I believe Funtoo uses this
for arch-agnostic packages.

Maybe rephrase this as being forbidden by Gentoo policy.

The rest of the patch looks ok.

Re: [gentoo-dev] [PATCH] skel.ebuild: Update comments for inherit, SLOT, KEYWORDS.

[Mart Raudsepp]

On Sun, 2017-12-31 at 14:31 +0100, Ulrich Müller wrote:
> -# DO NOT USE KEYWORDS="*".  This is deprecated and only for backward
> -# compatibility reasons.
> +# Do not use KEYWORDS="*".  This is not valid in an ebuild context.
>  KEYWORDS="~x86"

Maybe update to KEYWORDS="~amd64" while here?

Re: [gentoo-dev] [PATCH] skel.ebuild: Update comments for inherit, SLOT, KEYWORDS.

[Sergei Trofimovich]

[-- Attachment #1: Type: text/plain, Size: 1968 bytes --]

On Sun, 31 Dec 2017 14:31:00 +0100
Ulrich Müller <ulm@gentoo.org> wrote:

> epatch() is provided by epatch.eclass now. Also comment the inherit
> line, since not every ebuild will use it.
> 
> Empty SLOT doesn't disable slots, but is outright illegal in all EAPIs.
> Similar for KEYWORDS="*", which isn't "deprecated" but invalid.
> ---
>  skel.ebuild | 9 ++++-----
>  1 file changed, 4 insertions(+), 5 deletions(-)
> 
> diff --git a/skel.ebuild b/skel.ebuild
> index 7ac9dfb7d6d8..4c19b1de4cb8 100644
> --- a/skel.ebuild
> +++ b/skel.ebuild
> @@ -13,9 +13,9 @@
>  EAPI=6
>  
>  # inherit lists eclasses to inherit functions from. For example, an ebuild
> -# that needs the epatch function from eutils.eclass won't work without the
> +# that needs the epatch function from epatch.eclass won't work without the
>  # following line:
> -inherit eutils
> +#inherit epatch
>  #
>  # eclasses tend to list descriptions of how to use their functions properly.
>  # take a look at /usr/portage/eclass/ for more examples.
> @@ -46,7 +46,7 @@ LICENSE=""
>  # of each SLOT and remove everything else.
>  # Note that normal applications should use SLOT="0" if possible, since
>  # there should only be exactly one version installed at a time.
> -# DO NOT USE SLOT=""! This tells Portage to disable SLOTs for this package.
> +# Do not use SLOT="", because the SLOT variable must not be empty.
>  SLOT="0"
>  
>  # Using KEYWORDS, we can record masking information *inside* an ebuild
> @@ -63,8 +63,7 @@ SLOT="0"
>  # For binary packages, use -* and then list the archs the bin package
>  # exists for.  If the package was for an x86 binary package, then
>  # KEYWORDS would be set like this: KEYWORDS="-* x86"
> -# DO NOT USE KEYWORDS="*".  This is deprecated and only for backward
> -# compatibility reasons.
> +# Do not use KEYWORDS="*".  This is not valid in an ebuild context.
>  KEYWORDS="~x86"

Looks good.

-- 

  Sergei

[-- Attachment #2: Цифровая подпись OpenPGP --]
[-- Type: application/pgp-signature, Size: 195 bytes --]

Re: [gentoo-dev] [PATCH] skel.ebuild: Update comments for inherit, SLOT, KEYWORDS.

[Michał Górny]

W dniu nie, 31.12.2017 o godzinie 14∶31 +0100, użytkownik Ulrich Müller
napisał:
> epatch() is provided by epatch.eclass now. Also comment the inherit
> line, since not every ebuild will use it.
> 
> Empty SLOT doesn't disable slots, but is outright illegal in all EAPIs.
> Similar for KEYWORDS="*", which isn't "deprecated" but invalid.
> ---
>  skel.ebuild | 9 ++++-----
>  1 file changed, 4 insertions(+), 5 deletions(-)
> 
> diff --git a/skel.ebuild b/skel.ebuild
> index 7ac9dfb7d6d8..4c19b1de4cb8 100644
> --- a/skel.ebuild
> +++ b/skel.ebuild
> @@ -13,9 +13,9 @@
>  EAPI=6
>  
>  # inherit lists eclasses to inherit functions from. For example, an ebuild
> -# that needs the epatch function from eutils.eclass won't work without the
> +# that needs the epatch function from epatch.eclass won't work without the
>  # following line:
> -inherit eutils
> +#inherit epatch

Well, given that it's EAPI 6 and we do not want to encourage people to
use epatch there, I would prefer if we replaced that example with
something else. But it's just a wishful thinking here.

>  #
>  # eclasses tend to list descriptions of how to use their functions properly.
>  # take a look at /usr/portage/eclass/ for more examples.
> @@ -46,7 +46,7 @@ LICENSE=""
>  # of each SLOT and remove everything else.
>  # Note that normal applications should use SLOT="0" if possible, since
>  # there should only be exactly one version installed at a time.
> -# DO NOT USE SLOT=""! This tells Portage to disable SLOTs for this package.
> +# Do not use SLOT="", because the SLOT variable must not be empty.
>  SLOT="0"
>  
>  # Using KEYWORDS, we can record masking information *inside* an ebuild
> @@ -63,8 +63,7 @@ SLOT="0"
>  # For binary packages, use -* and then list the archs the bin package
>  # exists for.  If the package was for an x86 binary package, then
>  # KEYWORDS would be set like this: KEYWORDS="-* x86"
> -# DO NOT USE KEYWORDS="*".  This is deprecated and only for backward
> -# compatibility reasons.
> +# Do not use KEYWORDS="*".  This is not valid in an ebuild context.
>  KEYWORDS="~x86"
>  
>  # Comprehensive list of any and all USE flags leveraged in the ebuild,

LGTM.

-- 
Best regards,
Michał Górny

Re: [gentoo-dev] [PATCH] skel.ebuild: Update comments for inherit, SLOT, KEYWORDS.

[William Hubbs]

[-- Attachment #1: Type: text/plain, Size: 1297 bytes --]

On Sun, Dec 31, 2017 at 06:44:39PM +0100, Michał Górny wrote:
> W dniu nie, 31.12.2017 o godzinie 14∶31 +0100, użytkownik Ulrich Müller
> napisał:
> > epatch() is provided by epatch.eclass now. Also comment the inherit
> > line, since not every ebuild will use it.
> > 
> > Empty SLOT doesn't disable slots, but is outright illegal in all EAPIs.
> > Similar for KEYWORDS="*", which isn't "deprecated" but invalid.
> > ---
> >  skel.ebuild | 9 ++++-----
> >  1 file changed, 4 insertions(+), 5 deletions(-)
> > 
> > diff --git a/skel.ebuild b/skel.ebuild
> > index 7ac9dfb7d6d8..4c19b1de4cb8 100644
> > --- a/skel.ebuild
> > +++ b/skel.ebuild
> > @@ -13,9 +13,9 @@
> >  EAPI=6
> >  
> >  # inherit lists eclasses to inherit functions from. For example, an ebuild
> > -# that needs the epatch function from eutils.eclass won't work without the
> > +# that needs the epatch function from epatch.eclass won't work without the
> >  # following line:
> > -inherit eutils
> > +#inherit epatch
> 
> Well, given that it's EAPI 6 and we do not want to encourage people to
> use epatch there, I would prefer if we replaced that example with
> something else. But it's just a wishful thinking here.

+1, but I'm not going to suggest what to replace it with.

William

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 195 bytes --]

Re: [gentoo-dev] [PATCH] skel.ebuild: Update comments for inherit, SLOT, KEYWORDS.

[Pacho Ramos]

El dom, 31-12-2017 a las 12:12 -0600, William Hubbs escribió:
> On Sun, Dec 31, 2017 at 06:44:39PM +0100, Michał Górny wrote:
> > W dniu nie, 31.12.2017 o godzinie 14∶31 +0100, użytkownik Ulrich Müller
> > napisał:
> > > epatch() is provided by epatch.eclass now. Also comment the inherit
> > > line, since not every ebuild will use it.
> > > 
> > > Empty SLOT doesn't disable slots, but is outright illegal in all EAPIs.
> > > Similar for KEYWORDS="*", which isn't "deprecated" but invalid.
> > > ---
> > >  skel.ebuild | 9 ++++-----
> > >  1 file changed, 4 insertions(+), 5 deletions(-)
> > > 
> > > diff --git a/skel.ebuild b/skel.ebuild
> > > index 7ac9dfb7d6d8..4c19b1de4cb8 100644
> > > --- a/skel.ebuild
> > > +++ b/skel.ebuild
> > > @@ -13,9 +13,9 @@
> > >  EAPI=6
> > >  
> > >  # inherit lists eclasses to inherit functions from. For example, an
> > > ebuild
> > > -# that needs the epatch function from eutils.eclass won't work without
> > > the
> > > +# that needs the epatch function from epatch.eclass won't work without
> > > the
> > >  # following line:
> > > -inherit eutils
> > > +#inherit epatch
> > 
> > Well, given that it's EAPI 6 and we do not want to encourage people to
> > use epatch there, I would prefer if we replaced that example with
> > something else. But it's just a wishful thinking here.
> 
> +1, but I'm not going to suggest what to replace it with.
> 
> William

Why not eapply (and drop inherit then)? :/

Re: [gentoo-dev] [PATCH] skel.ebuild: Update comments for inherit, SLOT, KEYWORDS.

[William Hubbs]

[-- Attachment #1: Type: text/plain, Size: 1961 bytes --]

On Sun, Dec 31, 2017 at 08:07:33PM +0100, Pacho Ramos wrote:
> El dom, 31-12-2017 a las 12:12 -0600, William Hubbs escribió:
> > On Sun, Dec 31, 2017 at 06:44:39PM +0100, Michał Górny wrote:
> > > W dniu nie, 31.12.2017 o godzinie 14∶31 +0100, użytkownik Ulrich Müller
> > > napisał:
> > > > epatch() is provided by epatch.eclass now. Also comment the inherit
> > > > line, since not every ebuild will use it.
> > > > 
> > > > Empty SLOT doesn't disable slots, but is outright illegal in all EAPIs.
> > > > Similar for KEYWORDS="*", which isn't "deprecated" but invalid.
> > > > ---
> > > >  skel.ebuild | 9 ++++-----
> > > >  1 file changed, 4 insertions(+), 5 deletions(-)
> > > > 
> > > > diff --git a/skel.ebuild b/skel.ebuild
> > > > index 7ac9dfb7d6d8..4c19b1de4cb8 100644
> > > > --- a/skel.ebuild
> > > > +++ b/skel.ebuild
> > > > @@ -13,9 +13,9 @@
> > > >  EAPI=6
> > > >  
> > > >  # inherit lists eclasses to inherit functions from. For example, an
> > > > ebuild
> > > > -# that needs the epatch function from eutils.eclass won't work without
> > > > the
> > > > +# that needs the epatch function from epatch.eclass won't work without
> > > > the
> > > >  # following line:
> > > > -inherit eutils
> > > > +#inherit epatch
> > > 
> > > Well, given that it's EAPI 6 and we do not want to encourage people to
> > > use epatch there, I would prefer if we replaced that example with
> > > something else. But it's just a wishful thinking here.
> > 
> > +1, but I'm not going to suggest what to replace it with.
> > 
> > William
> 
> Why not eapply (and drop inherit then)? :/

Because that isn't the preferred way to apply patches since it requires
you to override src_prepare.

If the example was about how to apply patches, the correct way is to put
them in a PATCHES variable in global scope, but that's not the intent of
the example.

The intent is to show inheriting an eclass.

William

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 195 bytes --]

Re: [gentoo-dev] [PATCH] skel.ebuild: Update comments for inherit, SLOT, KEYWORDS.

[Ulrich Mueller]

[-- Attachment #1: Type: text/plain, Size: 368 bytes --]

>>>>> On Sun, 31 Dec 2017, Mart Raudsepp wrote:

> On Sun, 2017-12-31 at 14:31 +0100, Ulrich Müller wrote:
>> -# DO NOT USE KEYWORDS="*".  This is deprecated and only for backward
>> -# compatibility reasons.
>> +# Do not use KEYWORDS="*".  This is not valid in an ebuild context.
>>  KEYWORDS="~x86"

> Maybe update to KEYWORDS="~amd64" while here?

Done.

[-- Attachment #2: Type: application/pgp-signature, Size: 490 bytes --]

Re: [gentoo-dev] [PATCH] skel.ebuild: Update comments for inherit, SLOT, KEYWORDS.

[Ulrich Mueller]

[-- Attachment #1: Type: text/plain, Size: 550 bytes --]

>>>>> On Sun, 31 Dec 2017, Mike Gilbert wrote:

> On Sun, Dec 31, 2017 at 8:31 AM, Ulrich Müller <ulm@gentoo.org> wrote:
>> +# Do not use KEYWORDS="*".  This is not valid in an ebuild context.

> Portage actually does support KEYWORDS="*". I believe Funtoo uses
> this for arch-agnostic packages.

> Maybe rephrase this as being forbidden by Gentoo policy.

Not only by policy. It is not valid, according to PMS:
https://projects.gentoo.org/pms/6/pms.html#x1-280003.1.7
https://projects.gentoo.org/pms/6/pms.html#x1-710007.3.2

Ulrich

[-- Attachment #2: Type: application/pgp-signature, Size: 490 bytes --]

Re: [gentoo-dev] [PATCH] skel.ebuild: Update comments for inherit, SLOT, KEYWORDS.

[Ulrich Mueller]

[-- Attachment #1: Type: text/plain, Size: 860 bytes --]

>>>>> On Sun, 31 Dec 2017, William Hubbs wrote:

> On Sun, Dec 31, 2017 at 06:44:39PM +0100, Michał Górny wrote:
>> W dniu nie, 31.12.2017 o godzinie 14∶31 +0100, użytkownik Ulrich Müller
>> napisał:
>> > -# that needs the epatch function from eutils.eclass won't work without the
>> > +# that needs the epatch function from epatch.eclass won't work without the
>> >  # following line:
>> > -inherit eutils
>> > +#inherit epatch
>>
>> Well, given that it's EAPI 6 and we do not want to encourage people
>> to use epatch there, I would prefer if we replaced that example
>> with something else. But it's just a wishful thinking here.

> +1, but I'm not going to suggest what to replace it with.

How about one of these examples:
"eautoreconf function from autotools.eclass"
"tc-getCC function from toolchain-funcs.eclass"

Ulrich

[-- Attachment #2: Type: application/pgp-signature, Size: 490 bytes --]

Re: [gentoo-dev] [PATCH] skel.ebuild: Update comments for inherit, SLOT, KEYWORDS.

[M. J. Everitt]

[-- Attachment #1.1: Type: text/plain, Size: 420 bytes --]

On 31/12/17 19:59, Ulrich Mueller wrote:
>
>> +1, but I'm not going to suggest what to replace it with.
> How about one of these examples:
> "eautoreconf function from autotools.eclass"
> "tc-getCC function from toolchain-funcs.eclass"
>
> Ulrich
I second eautoreconf (with mention of eapply_user and/or default) as I
recently mistakenly put it in the wrong phase function .... [d'oh, the
n00bness!]

MJE


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 862 bytes --]

[gentoo-dev] [PATCH v2] skel.ebuild: Update comments for inherit, SLOT, KEYWORDS.

[Ulrich Müller]

[-- Attachment #1: Type: text/plain, Size: 3378 bytes --]

Change the example used for inherit. Also comment the inherit line,
since not every ebuild will use it.

Empty SLOT doesn't disable slots, but is outright illegal in all EAPIs.
Similar for KEYWORDS="*", which isn't "deprecated" but invalid.

Replace x86 by amd64 as an example, where appropriate.
---

Changes in v2:
- inherit uses autotools rather than epatch as an example.
- x86 replaced by amd64 where appropriate.
- More accurate wording for IUSE.

 skel.ebuild | 27 +++++++++++++--------------
 1 file changed, 13 insertions(+), 14 deletions(-)

diff --git a/skel.ebuild b/skel.ebuild
index 591b58e8d367..129f61dcdb29 100644
--- a/skel.ebuild
+++ b/skel.ebuild
@@ -13,9 +13,9 @@
 EAPI=6
 
 # inherit lists eclasses to inherit functions from. For example, an ebuild
-# that needs the epatch function from eutils.eclass won't work without the
-# following line:
-inherit eutils
+# that needs the eautoreconf function from autotools.eclass won't work
+# without the following line:
+#inherit autotools
 #
 # eclasses tend to list descriptions of how to use their functions properly.
 # take a look at /usr/portage/eclass/ for more examples.
@@ -46,7 +46,7 @@ LICENSE=""
 # of each SLOT and remove everything else.
 # Note that normal applications should use SLOT="0" if possible, since
 # there should only be exactly one version installed at a time.
-# DO NOT USE SLOT=""! This tells Portage to disable SLOTs for this package.
+# Do not use SLOT="", because the SLOT variable must not be empty.
 SLOT="0"
 
 # Using KEYWORDS, we can record masking information *inside* an ebuild
@@ -54,22 +54,21 @@ SLOT="0"
 # set the KEYWORDS variable for every ebuild so that it contains the names of
 # all the architectures with which the ebuild works.  All of the official
 # architectures can be found in the arch.list file which is in
-# /usr/portage/profiles/.  Usually you should just set this to "~x86".  The ~
-# in front of the architecture indicates that the package is new and should be
-# considered unstable until testing proves its stability.  So, if you've
-# confirmed that your ebuild works on x86 and ppc, you'd specify:
-# KEYWORDS="~x86 ~ppc"
+# /usr/portage/profiles/.  Usually you should just set this to "~amd64".
+# The ~ in front of the architecture indicates that the package is new and
+# should be considered unstable until testing proves its stability.  So, if
+# you've confirmed that your ebuild works on amd64 and ppc, you'd specify:
+# KEYWORDS="~amd64 ~ppc"
 # Once packages go stable, the ~ prefix is removed.
 # For binary packages, use -* and then list the archs the bin package
 # exists for.  If the package was for an x86 binary package, then
 # KEYWORDS would be set like this: KEYWORDS="-* x86"
-# DO NOT USE KEYWORDS="*".  This is deprecated and only for backward
-# compatibility reasons.
-KEYWORDS="~x86"
+# Do not use KEYWORDS="*"; this is not valid in an ebuild context.
+KEYWORDS="~amd64"
 
 # Comprehensive list of any and all USE flags leveraged in the ebuild,
-# with the exception of any ARCH specific flags, i.e. "ppc", "sparc",
-# "x86" and "alpha".  Not needed if the ebuild doesn't use any USE flags.
+# with some exceptions, e.g., ARCH specific flags like "amd64" or "ppc".
+# Not needed if the ebuild doesn't use any USE flags.
 IUSE="gnome X"
 
 # A space delimited list of portage features to restrict. man 5 ebuild
-- 
2.15.1

[-- Attachment #2: Type: application/pgp-signature, Size: 490 bytes --]