Skip to contentSkip to conversation

go-vendor.bbclass: Unpack into BP for destsuffix

PATCH Superseded John Ripple 3 participants 5 messages Jun 18

v1 of 3 Superseded Latest is v3 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 go-vendor.bbclass: Unpack into BP for destsuffix

The go-vendor class has a hardcoded destsuffix variable used to unpack sources fetched with the go_src_uri function. Change this unpack location to take the base package name into account (BP) as is the new standard.

---
 meta/classes/go-vendor.bbclass | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
meta/classes/go-vendor.bbclass +3 -2
index 6ec6178add..f854d14bdf 100644
@@ -19,7 +19,7 @@ inherit go-mod
 def go_src_uri(repo, version, path=None, subdir=None, \
                 vcs='git', replaces=None, pathmajor=None):
 
-    destsuffix = "git/src/import/vendor.fetch"
+    destsuffix = "${BP}/src/import/vendor.fetch"
     module_path = repo if not path else path
 
     src_uri = "{}://{};name={}".format(vcs, repo, module_path.replace('/', '.'))
@@ -58,7 +58,8 @@ python do_go_vendor() {
     if not src_uri:
         bb.fatal("SRC_URI is empty")
 
-    default_destsuffix = "git/src/import/vendor.fetch"
+    base_package = d.getVar('BP')
+    default_destsuffix = "{}/src/import/vendor.fetch".format(base_package)
     fetcher = bb.fetch2.Fetch(src_uri, d)
     go_import = d.getVar('GO_IMPORT')
     source_dir = d.getVar('S')
#1 lore raw
Quoted 5 lines · John Ripple via lists.openembedded.org

The go-vendor class has a hardcoded destsuffix variable used to unpack sources fetched with the go_src_uri function. Change this unpack location to take the base package name into account (BP) as is the new standard.

I have no particular issues with the change, but the commit log doesn't tell us why this is needed. Was something breaking ? is there any risk to the assumptions that other classes/recipes/layers may have made about the location of the unpacked files ?

Bruce

Quoted 43 lines
---
 meta/classes/go-vendor.bbclass | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
meta/classes/go-vendor.bbclass +1 -1
index 6ec6178add..f854d14bdf 100644
@@ -19,7 +19,7 @@ inherit go-mod
 def go_src_uri(repo, version, path=None, subdir=None, \
                 vcs='git', replaces=None, pathmajor=None):
 
-    destsuffix = "git/src/import/vendor.fetch"
+    destsuffix = "${BP}/src/import/vendor.fetch"
     module_path = repo if not path else path
 
     src_uri = "{}://{};name={}".format(vcs, repo,

module_path.replace('/', '.'))

meta/classes/go-vendor.bbclass +2 -2
@@ -58,7 +58,8 @@ python do_go_vendor() {
     if not src_uri:
         bb.fatal("SRC_URI is empty")
 
-    default_destsuffix = "git/src/import/vendor.fetch"
+    base_package = d.getVar('BP')
+    default_destsuffix = "{}/src/import/vendor.fetch".format(base_package)
     fetcher = bb.fetch2.Fetch(src_uri, d)
     go_import = d.getVar('GO_IMPORT')
     source_dir = d.getVar('S')
 
-=-=-=-=-=-=-=-=-=-=-=-

Links: You receive all messages sent to this group. View/Reply Online (#239109):
https://lists.openembedded.org/g/openembedded-core/message/239109
Mute This Topic: https://lists.openembedded.org/mt/119874754/1050810
Group Owner: [email protected]
Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub [ [email protected]]
-=-=-=-=-=-=-=-=-=-=-=-

Signature · 3 lines
- Thou shalt not follow the NULL pointer, for chaos and madness await thee
at its end
- "Use the force Harry" - Gandalf, Star Trek II
#2 lore raw
Quoted 6 lines

I have no particular issues with the change, but the commit log doesn't tell us why this is needed. Was something breaking ? is there any risk to the assumptions that other classes/recipes/layers may have made about the location of the unpacked files ?

The go-vendor class was unpacking dependencies to ${UNPACKDIR}/git
and then needed S = "${UNPACKDIR}/git" to find the dependencies.
Changing the S variable caused the do_qa_unpack() ( https://git.openembedded.org/openembedded-core/tree/meta/classes-global/insane.bbclass#n1450 ) from insane.bbclass to throw an error.

There shouldn't be any other assumptions from other classes/recipes/layers that have been broken. The recipes I tested this change against used sources fetched using the go_src_uri() function, regular git fetching, and local files. There weren't any issues with moving to the go dependencies to the new location. As far as I know nothing in openembedded-core uses the go-vendor class making it trickier to test against a known baseline.

John

#3 lore raw
Quoted 7 lines · John Ripple via lists.openembedded.org
Quoted 6 lines

I have no particular issues with the change, but the commit log doesn't tell us why this is needed. Was something breaking ? is there any risk to the assumptions that other classes/recipes/layers may have made about the location of the unpacked files ?

Quoted 11 lines

The go-vendor class was unpacking dependencies to ${UNPACKDIR}/git and then needed S = "${UNPACKDIR}/git" to find the dependencies. Changing the S variable caused the do_qa_unpack() from insane.bbclass to throw an error.

There shouldn't be any other assumptions from other classes/recipes/layers that have been broken. The recipes I tested this change against used sources fetched using the go_src_uri() function, regular git fetching, and local files. There weren't any issues with moving to the go dependencies to the new location. As far as I know nothing in openembedded-core uses the go-vendor class making it trickier to test against a known baseline.

It does sound a lot like we're missing some test cases as if I understand it correctly, that class can't have worked for a while?

Cheers,

Richard

#4 lore raw
Quoted 4 lines

It does sound a lot like we're missing some test cases as if I understand it correctly, that class can't have worked for a while?

I've used the go-vendor class on Scarthgap and have only started to look at it on Wrynose with the recent release. It would have thrown an error since whenever the insane.bbclass explicitly denied setting S = "${UNPACKDIR}/git". It's probably slipped under the radar because nothing uses it in the core layers.

John