Skip to contentSkip to conversation

rsnapshot: make this an allarch package

PATCH NAK André Draszik 3 participants 6 messages Jul 24, 2015 – Sep 23, 2015 1 message not in the archive

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 rsnapshot: make this an allarch package

From: André Draszik <[email protected]>

rsnapshot is just some perl script(s), no machine specific binaries.

---
 meta-oe/recipes-support/rsnapshot/rsnapshot_git.bb | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
meta-oe/recipes-support/rsnapshot/rsnapshot_git.bb +1 -1
index d680511..0593eef 100644
@@ -29,7 +29,7 @@ SRC_URI = "git://github.com/DrHyde/${BPN};branch=master;protocol=git \
 
 S = "${WORKDIR}/git"
 
-inherit autotools
+inherit allarch autotools
 
 # Fix rsnapshot.conf.default:
 # don't inject the host path into target configs.
Signature · 1 line
2.1.4
#1 lore raw
Quoted 3 lines · André Draszik

From: André Draszik <[email protected]>

rsnapshot is just some perl script(s), no machine specific binaries.

NAK

The recipe has bunch of RDEPENDS on TUNE_PKGARCH recipes so it cannot be allarch.

Recipe by default is TUNE_PKGARCH not MACHINE_ARCH as commit message may suggest.

Quoted 25 lines
---
 meta-oe/recipes-support/rsnapshot/rsnapshot_git.bb | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
meta-oe/recipes-support/rsnapshot/rsnapshot_git.bb +1 -1
index d680511..0593eef 100644
@@ -29,7 +29,7 @@ SRC_URI = "git://github.com/DrHyde/${BPN};branch=master;protocol=git \
 
 S = "${WORKDIR}/git"
 
-inherit autotools
+inherit allarch autotools
 
 # Fix rsnapshot.conf.default:
 # don't inject the host path into target configs.
Signature · 7 lines
2.1.4

-- 
_______________________________________________
Openembedded-devel mailing list
[email protected]
http://lists.openembedded.org/mailman/listinfo/openembedded-devel
Signature · 1 line
Martin 'JaMa' Jansa     jabber: [email protected]
#2 lore raw

PATCH 1/1 ref-manual: clarify allarch wrt RDPENDS and TUNE_PKGARCH

From: André Draszik <[email protected]>

See e.g.
http://permalink.gmane.org/gmane.comp.handhelds.openembedded/64067

---
 documentation/ref-manual/ref-classes.xml | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)
documentation/ref-manual/ref-classes.xml +16 -0
index d482ca0..a9b2f3c 100644
@@ -51,6 +51,22 @@
         that produce executable binaries (such as building the cross-compiler
         and a C library as pre-requisites, and splitting out of debug symbols
         during packaging).
+        <note>
+            Unlike e.g. Debian, OpenEmbedded recipes that produce packages
+            which
+            <link linkend='var-RDEPENDS'><filename>RDEPENDS</filename></link>
+            on
+            <link linkend='var-TUNE_PKGARCH'><filename>TUNE_PKGARCH</filename></link>
+            packages should never be made <filename>allarch</filename>, even
+            if they do not produce architecture-specific output. This would
+            cause the do_package_write_* tasks to have different signatures
+            for
+            <link linkend='var-MACHINE'><filename>MACHINE</filename></link>s
+            with different
+            <link linkend='var-TUNE_PKGARCH'><filename>TUNE_PKGARCH</filename></link>,
+            thus unnecessary rebuilds every single time an image for a different
+            MACHINE is built (even without any change to the recipe).
+        </note>
     </para>
 
     <para>
Signature · 1 line
2.1.4
#3 lore raw
In reply to Martin Jansa
Quoted 8 lines · Martin Jansa
Quoted 3 lines · André Draszik

From: André Draszik <[email protected]>

rsnapshot is just some perl script(s), no machine specific binaries.

Quoted 7 lines

NAK

The recipe has bunch of RDEPENDS on TUNE_PKGARCH recipes so it cannot be allarch.

Recipe by default is TUNE_PKGARCH not MACHINE_ARCH as commit message may suggest.

I understand the NAK since it has runtime dependencies on binary compiled programs like rsync which are architecture dependent due to different ISA and so on. However, from another perspective, its a script which is invoking/executing a standard utility much like its own interpreter say bash or perl whichever it needs and bash, perl etc. are natively compiled binaries. going by this logic we can really not have allarch package at all. So IMO runtime dependencies
for script based packages should not bar it from being a allarch package. Since, I can still take the same ipk and install it on arm or mips and it will work as long as they provide its dependencies like rsync perl etc.

Quoted 34 lines
Quoted 21 lines

---
meta-oe/recipes-support/rsnapshot/rsnapshot_git.bb | 2 +-

1 file changed, 1 insertion(+), 1 deletion(-)
meta-oe/recipes-support/rsnapshot/rsnapshot_git.bb +0 -0
index d680511..0593eef 100644
@@ -29,7 +29,7 @@ SRC_URI = "git://github.com/DrHyde/${BPN};branch=master;protocol=git \
 

S = "${WORKDIR}/git"

meta-oe/recipes-support/rsnapshot/rsnapshot_git.bb +1 -1
-inherit autotools
+inherit allarch autotools

# Fix rsnapshot.conf.default:
# don't inject the host path into target configs.
--
2.1.4

Quoted 5 lines

--
_______________________________________________
Openembedded-devel mailing list
[email protected]
http://lists.openembedded.org/mailman/listinfo/openembedded-devel

--
Martin 'JaMa' Jansa jabber: [email protected]
--
_______________________________________________
Openembedded-devel mailing list
[email protected]
http://lists.openembedded.org/mailman/listinfo/openembedded-devel

#4 lore raw
Quoted 17 lines · Khem Raj
Quoted 8 lines · Martin Jansa
Quoted 3 lines · André Draszik

From: André Draszik <[email protected]>

rsnapshot is just some perl script(s), no machine specific binaries.

Quoted 7 lines

NAK

The recipe has bunch of RDEPENDS on TUNE_PKGARCH recipes so it cannot be allarch.

Recipe by default is TUNE_PKGARCH not MACHINE_ARCH as commit message may suggest.

Quoted 6 lines

I understand the NAK since it has runtime dependencies on binary compiled programs like rsync which are architecture dependent due to different ISA and so on. However, from another perspective, its a script which is invoking/executing a standard utility much like its own interpreter say bash or perl whichever it needs and bash, perl etc. are natively compiled binaries. going by this logic we can really not have allarch package at all. So IMO runtime dependencies
for script based packages should not bar it from being a allarch package. Since, I can still take the same ipk and install it on arm or mips and it will work as long as they provide its dependencies like rsync perl etc.

With current sstate checksum generators accepting this will cause all these "allarch" recipes to be "rebuilt" (from sstate) every time you change MACHINE to something with different TUNE_PKGARCH.

Keeping it TUNE_PKGARCH means that you'll always get exactly the same package from feed (not the last one "built" randomly changing when builder builds different MACHINEs in a loop).

Quoted 44 lines
Quoted 35 lines
Quoted 22 lines

---
meta-oe/recipes-support/rsnapshot/rsnapshot_git.bb | 2 +-

1 file changed, 1 insertion(+), 1 deletion(-)
meta-oe/recipes-support/rsnapshot/rsnapshot_git.bb +0 -0
index d680511..0593eef 100644
@@ -29,7 +29,7 @@ SRC_URI = "git://github.com/DrHyde/${BPN};branch=master;protocol=git \
 

S = "${WORKDIR}/git"

meta-oe/recipes-support/rsnapshot/rsnapshot_git.bb +1 -1
-inherit autotools
+inherit allarch autotools

# Fix rsnapshot.conf.default:
# don't inject the host path into target configs.
--
2.1.4

Quoted 5 lines

--
_______________________________________________
Openembedded-devel mailing list
[email protected]
http://lists.openembedded.org/mailman/listinfo/openembedded-devel

--
Martin 'JaMa' Jansa jabber: [email protected]
--
_______________________________________________
Openembedded-devel mailing list
[email protected]
http://lists.openembedded.org/mailman/listinfo/openembedded-devel

Signature · 4 lines
_______________________________________________
Openembedded-devel mailing list
[email protected]
http://lists.openembedded.org/mailman/listinfo/openembedded-devel
Signature · 1 line
Martin 'JaMa' Jansa     jabber: [email protected]

A message that is not in the archive (sent off-list or not yet received): [email protected]

#6 lore raw

Re: [yocto-docs][PATCH] ref-manual: clarify allarch wrt RDPENDS and TUNE_PKGARCH

Hi,

Is there anything wrong with this patch?

Cheers,
Andre'

Quoted 40 lines · On 28 July 2015 at 09:04, <[email protected]> wrote:

From: André Draszik <[email protected]>

See e.g.
http://permalink.gmane.org/gmane.comp.handhelds.openembedded/64067

---
 documentation/ref-manual/ref-classes.xml | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)
documentation/ref-manual/ref-classes.xml +16 -1
index d482ca0..a9b2f3c 100644
@@ -51,6 +51,22 @@
         that produce executable binaries (such as building the cross-compiler
         and a C library as pre-requisites, and splitting out of debug symbols
         during packaging).
+        <note>
+            Unlike e.g. Debian, OpenEmbedded recipes that produce packages
+            which
+            <link linkend='var-RDEPENDS'><filename>RDEPENDS</filename></link>
+            on
+            <link linkend='var-TUNE_PKGARCH'><filename>TUNE_PKGARCH</filename></link>
+            packages should never be made <filename>allarch</filename>, even
+            if they do not produce architecture-specific output. This would
+            cause the do_package_write_* tasks to have different signatures
+            for
+            <link linkend='var-MACHINE'><filename>MACHINE</filename></link>s
+            with different
+            <link linkend='var-TUNE_PKGARCH'><filename>TUNE_PKGARCH</filename></link>,
+            thus unnecessary rebuilds every single time an image for a different
+            MACHINE is built (even without any change to the recipe).
+        </note>
     </para>
 
     <para>
--

2.1.4