Skip to contentSkip to conversation

numerous questions about some ref manual variable glossary entries

DISCUSSION Robert P. J. Day 2 participants 4 messages Nov 3, 2013 – Nov 4, 2013

lore
Share & tools
lore
#0 lore raw

wasn't sure what to do about some of these so i'd appreciate some
clarification. and away we go, mostly in alphabetical order ...

http://www.yoctoproject.org/docs/latest/ref-manual/ref-manual.html#var-BBLAYERS_NON_REMOVABLE

that entry *really* implies you need the meta-yocto and
meta-yocto-bsp layers for a successful build, which clearly isn't true as lots of layers work with simply oe-core. can/should that explanation be reworded to not be quite so emphatic?

http://www.yoctoproject.org/docs/latest/ref-manual/ref-manual.html#var-INITSCRIPT_NAME

reference to ${etcdir} ... how exactly does that differ from
${sysconfdir}? i've always understood that developers should use ${sysconfdir} in their recipe files.

http://www.yoctoproject.org/docs/latest/ref-manual/ref-manual.html#var-NO_RECOMMENDATIONS
http://www.yoctoproject.org/docs/latest/ref-manual/ref-manual.html#var-PACKAGE_EXCLUDE

both of those claim no support for DEB packaging, is that still
true? i still don't know enough python to check the actual code but i'm getting there.

http://www.yoctoproject.org/docs/latest/ref-manual/ref-manual.html#var-RPROVIDES

claims that a package's own name is already implicitly provided in
its RPROVIDES list. doesn't seem that way -- it's true for PROVIDES, but doesn't seem true for RPROVIDES. thoughts?

i think that might be it.

rday

Signature · 7 lines
========================================================================
Robert P. J. Day                                 Ottawa, Ontario, CANADA
                        http://crashcourse.ca

Twitter:                                       http://twitter.com/rpjday
LinkedIn:                               http://ca.linkedin.com/in/rpjday
========================================================================
#1 lore raw

Hi Robert,

Quoted 10 lines · On Sunday 03 November 2013 05:53:54 Robert P. J. Day wrote:

wasn't sure what to do about some of these so i'd appreciate some
clarification. and away we go, mostly in alphabetical order ...

http://www.yoctoproject.org/docs/latest/ref-manual/ref-manual.html#var-BBLAY
ERS_NON_REMOVABLE

that entry *really* implies you need the meta-yocto and
meta-yocto-bsp layers for a successful build, which clearly isn't true as lots of layers work with simply oe-core. can/should that explanation be reworded to not be quite so emphatic?

Yes, agreed. The entry should mention that this variable is only actually used by Hob as well.

Quoted 6 lines

http://www.yoctoproject.org/docs/latest/ref-manual/ref-manual.html#var-INITS
CRIPT_NAME

reference to ${etcdir} ... how exactly does that differ from
${sysconfdir}? i've always understood that developers should use ${sysconfdir} in their recipe files.

I don't know where this came from but it is wrong. It should say ${sysconfdir} instead.

Quoted 8 lines

http://www.yoctoproject.org/docs/latest/ref-manual/ref-manual.html#var-NO_RE
COMMENDATIONS
http://www.yoctoproject.org/docs/latest/ref-manual/ref-manual.html#var-PACK
AGE_EXCLUDE

both of those claim no support for DEB packaging, is that still
true? i still don't know enough python to check the actual code but i'm getting there.

Yes, these were only introduced recently and AFAIK this is still the case.

Quoted 6 lines

http://www.yoctoproject.org/docs/latest/ref-manual/ref-manual.html#var-RPROV
IDES

claims that a package's own name is already implicitly provided in
its RPROVIDES list. doesn't seem that way -- it's true for PROVIDES, but doesn't seem true for RPROVIDES. thoughts?

It was me who got Scott to add this - I did so after seeing a number of people trying to do RPROVIDES_${PN} = "recipename" in the hope of fixing some problem they were having; but this won't do anything. What makes you think it isn't the case?

Cheers,
Paul

Signature · 2 lines
Paul Eggleton
Intel Open Source Technology Centre
#2 lore raw
Quoted 10 lines · Paul Eggleton

Hi Robert,

Quoted 7 lines · On Sunday 03 November 2013 05:53:54 Robert P. J. Day wrote:

wasn't sure what to do about some of these so i'd appreciate
some clarification. and away we go, mostly in alphabetical order
...

http://www.yoctoproject.org/docs/latest/ref-manual/ref-manual.html#var-BBLAY
ERS_NON_REMOVABLE

Quoted 7 lines
Quoted 4 lines

that entry *really* implies you need the meta-yocto and
meta-yocto-bsp layers for a successful build, which clearly isn't true as lots of layers work with simply oe-core. can/should that explanation be reworded to not be quite so emphatic?

Yes, agreed. The entry should mention that this variable is only actually used by Hob as well.

at the risk of sounding lazy, i'll let someone more comfortable with
this topic tweak that text.

Quoted 9 lines
Quoted 6 lines

http://www.yoctoproject.org/docs/latest/ref-manual/ref-manual.html#var-INITS
CRIPT_NAME

reference to ${etcdir} ... how exactly does that differ from
${sysconfdir}? i've always understood that developers should use ${sysconfdir} in their recipe files.

I don't know where this came from but it is wrong. It should say ${sysconfdir} instead.

there's more to it than that -- there are a *number* of references
to "etcdir" scattered across a number of layers:

meta-angstrom/recipes-angstrom/angstrom/angstrom-feed-configs.bb: etcdir = bb.data.expand('${sysconfdir}/opkg', d)
meta-angstrom/recipes-angstrom/angstrom/angstrom-feed-configs.bb: do_split_packages(d, etcdir, '^locale-(.*)\.conf$', 'angstrom-locale-%s-config', 'Angstrom feed config for the %s locale', extra_depends='', allow_links=True)
meta-openembedded/meta-oe/recipes-extended/zsh/zsh.inc: --enable-etcdir=${sysconfdir} \
oe-core/meta/conf/documentation.conf:INITSCRIPT_NAME[doc] = "The filename of the initscript as installed to ${etcdir}/init.d. The variable is mandatory and is used in recipes when using update-rc.d.bbclass."
oe-core/meta/recipes-devtools/quilt/quilt/install.patch:-etcdir := $(subst /usr/etc,/etc,$(prefix)/etc)
oe-core/meta/recipes-devtools/quilt/quilt/install.patch:+etcdir := @sysconfdir@
oe-core/meta/recipes-bsp/pcmciautils/pcmciautils.inc:export etcdir = "${sysconfdir}"
oe-core/scripts/lib/mic/bootstrap.py: etcdir = _path('/etc')
oe-core/scripts/lib/mic/bootstrap.py: if not os.path.exists(etcdir):
oe-core/scripts/lib/mic/bootstrap.py: os.makedirs(etcdir)
openembedded-core-contrib/meta/recipes-devtools/quilt/quilt/install.patch:-etcdir := $(subst /usr/etc,/etc,$(prefix)/etc)
openembedded-core-contrib/meta/recipes-devtools/quilt/quilt/install.patch:+etcdir := @sysconfdir@
openembedded-core-contrib/meta/recipes-bsp/pcmciautils/pcmciautils.inc:export etcdir = "${sysconfdir}"
yocto-docs/documentation/ref-manual/ref-variables.xml: The filename of the initscript as installed to <filename>${etcdir}/init.d</filename>.

so it's not like that's an isolated occurrence. i have no idea *what* to make of the above. is it all correct other than that single reference above?

Quoted 11 lines
Quoted 6 lines

http://www.yoctoproject.org/docs/latest/ref-manual/ref-manual.html#var-RPROV
IDES

claims that a package's own name is already implicitly provided in
its RPROVIDES list. doesn't seem that way -- it's true for PROVIDES, but doesn't seem true for RPROVIDES. thoughts?

It was me who got Scott to add this - I did so after seeing a number of people trying to do RPROVIDES_${PN} = "recipename" in the hope of fixing some problem they were having; but this won't do anything. What makes you think it isn't the case?

i played with "bb show" to get the following (for the
arbitrarily-chosen recipe zlib):

$ bb show -r zlib PROVIDES
Parsing recipes..done.
# PROVIDES=${PN}
PROVIDES="zlib "
$ bb show -r zlib RPROVIDES
Parsing recipes..done.
RPROVIDES=""
$

where you can see that, for the example zlib recipe, PROVIDES contains the recipe name but RPROVIDES does not, so i assumed RPROVIDES didn't automatically provide the package's own name. am i misreading what i'm seeing above?

rday

Signature · 7 lines
========================================================================
Robert P. J. Day                                 Ottawa, Ontario, CANADA
                        http://crashcourse.ca

Twitter:                                       http://twitter.com/rpjday
LinkedIn:                               http://ca.linkedin.com/in/rpjday
========================================================================
#3 lore raw
Quoted 44 lines · On Monday 04 November 2013 10:40:45 Robert P. J. Day wrote:
Quoted 10 lines · Paul Eggleton
Quoted 6 lines · On Sunday 03 November 2013 05:53:54 Robert P. J. Day wrote:

http://www.yoctoproject.org/docs/latest/ref-manual/ref-manual.html#var-I
NITS CRIPT_NAME

reference to ${etcdir} ... how exactly does that differ from
${sysconfdir}? i've always understood that developers should use ${sysconfdir} in their recipe files.

I don't know where this came from but it is wrong. It should say ${sysconfdir} instead.

there's more to it than that -- there are a *number* of references
to "etcdir" scattered across a number of layers:

meta-angstrom/recipes-angstrom/angstrom/angstrom-feed-configs.bb: etcdir
= bb.data.expand('${sysconfdir}/opkg', d) meta-angstrom/recipes-angstrom/angstrom/angstrom-feed-configs.bb:
do_split_packages(d, etcdir, '^locale-(.*)\.conf$', 'angstrom-locale-%s-config', 'Angstrom feed config for the %s locale', extra_depends='', allow_links=True) meta-openembedded/meta-oe/recipes-extended/zsh/zsh.inc:
--enable-etcdir=${sysconfdir} \
oe-core/meta/conf/documentation.conf:INITSCRIPT_NAME[doc] = "The filename of the initscript as installed to ${etcdir}/init.d. The variable is mandatory and is used in recipes when using update-rc.d.bbclass."
oe-core/meta/recipes-devtools/quilt/quilt/install.patch:-etcdir := $(subst
/usr/etc,/etc,$(prefix)/etc) oe-core/meta/recipes-devtools/quilt/quilt/install.patch:+etcdir
:= @sysconfdir@ oe-core/meta/recipes-bsp/pcmciautils/pcmciautils.inc:export
etcdir = "${sysconfdir}" oe-core/scripts/lib/mic/bootstrap.py:
etcdir = _path('/etc') oe-core/scripts/lib/mic/bootstrap.py: if
not os.path.exists(etcdir): oe-core/scripts/lib/mic/bootstrap.py:
os.makedirs(etcdir)
openembedded-core-contrib/meta/recipes-devtools/quilt/quilt/install.patch:-
etcdir := $(subst /usr/etc,/etc,$(prefix)/etc)
openembedded-core-contrib/meta/recipes-devtools/quilt/quilt/install.patch:+
etcdir := @sysconfdir@
openembedded-core-contrib/meta/recipes-bsp/pcmciautils/pcmciautils.inc:expo rt etcdir = "${sysconfdir}" yocto-docs/documentation/ref-manual/ref-variables.xml:
The filename of the initscript as installed to <filename>${etcdir}/init.d</filename>.

Quoted 3 lines

so it's not like that's an isolated occurrence. i have no idea *what* to make of the above. is it all correct other than that single reference above?

None of those are trying to access an "etcdir" variable defined system-wide. There is no such variable, and most of the examples are unrelated to eachother.

Quoted 25 lines
Quoted 5 lines
Quoted 5 lines

http://www.yoctoproject.org/docs/latest/ref-manual/ref-manual.html#var-R
PROV IDES

claims that a package's own name is already implicitly provided in

Quoted 7 lines
Quoted 2 lines

its RPROVIDES list. doesn't seem that way -- it's true for PROVIDES, but doesn't seem true for RPROVIDES. thoughts?

It was me who got Scott to add this - I did so after seeing a number of people trying to do RPROVIDES_${PN} = "recipename" in the hope of fixing some problem they were having; but this won't do anything. What makes you think it isn't the case?

i played with "bb show" to get the following (for the
arbitrarily-chosen recipe zlib):

$ bb show -r zlib PROVIDES
Parsing recipes..done.
# PROVIDES=${PN}
PROVIDES="zlib "
$ bb show -r zlib RPROVIDES
Parsing recipes..done.
RPROVIDES=""
$
Quoted 4 lines

where you can see that, for the example zlib recipe, PROVIDES contains the recipe name but RPROVIDES does not, so i assumed RPROVIDES didn't automatically provide the package's own name. am i misreading what i'm seeing above?

This is what I meant by "implicit". That doesn't mean printing out the variable will show that it explicitly contains the value.

Cheers,
Paul

Signature · 2 lines
Paul Eggleton
Intel Open Source Technology Centre