Skip to contentSkip to conversation

CI: select CC based on CC_PACKAGE (again)

PATCH Superseded Ævar Arnfjörð Bjarmason 5 participants 18 messages Apr 21, 2022 – Apr 22, 2022

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 CI: select CC based on CC_PACKAGE (again)

Fix a regression in 707d2f2fe86 (CI: use "$runs_on_pool", not "$jobname" to select packages & config, 2021-11-23).

In that commit I changed CC=gcc from CC=gcc-9, but on OSX the "gcc" in $PATH points to clang, we need to use gcc-9 instead. Likewise for the linux-gcc job CC=gcc-8 was changed to the implicit CC=gcc, which would select GCC 9.4.0 instead of GCC 8.4.0.

Furthermore in 25715419bf4 (CI: don't run "make test" twice in one job, 2021-11-23) when the "linux-TEST-vars" job was split off from "linux-gcc" the "cc_package: gcc-8" line was copied along with it.

That wasn't a bug per-se, as that "make test" would have run under GCC 8 before the split into two jobs, but the point of selecting different compiler for these jobs is to get better coverage, and to narrow down any issues with a given compiler to the job that runs it. Since the "linux-TEST-vars" job is already special in other ways (in running with various GIT_TEST_* variables), and we've got the "linux-gcc" job covering gcc-8 let's have it used the default system compiler instead.

---
Range-diff:

1:  94abb826627 < -:  ----------- CI: have osx-gcc use gcc, not clang
-:  ----------- > 1:  d89ad4d5b7c CI: select CC based on CC_PACKAGE (again)

The range-diff is against a relevant patch in v4 of a larger CI series of mine[1]. This fix is independent of that, and on top of master. See
[2] for the original fix and discussion.

As noted in the updated commit message not only the OSX jobs were affected, but also the linux-gcc job. This fixes both, along with a small fix to the related linux-TEST-vars recipe.

1. https://lore.kernel.org/git/[email protected]/
2. https://lore.kernel.org/git/[email protected]/

 .github/workflows/main.yml | 1 -
 ci/lib.sh                  | 2 +-
 2 files changed, 1 insertion(+), 2 deletions(-)
.github/workflows/main.yml +0 -1
index c35200defb9..f12819a00d7 100644
@@ -236,7 +236,6 @@ jobs:
           - jobname: linux-TEST-vars
             cc: gcc
             os: ubuntu
-            cc_package: gcc-8
             pool: ubuntu-latest
           - jobname: osx-clang
             cc: clang
ci/lib.sh +1 -1
index cbc2f8f1caa..44007dcf93b 100755
@@ -122,7 +122,7 @@ then
 	test macos != "$CI_OS_NAME" || CI_OS_NAME=osx
 	CI_REPO_SLUG="$GITHUB_REPOSITORY"
 	CI_JOB_ID="$GITHUB_RUN_ID"
-	CC="${CC:-gcc}"
+	CC="${CC:-${CC_PACKAGE:-gcc}}"
 	DONT_SKIP_TAGS=t
 
 	cache_dir="$HOME/none"
Signature · 1 line
2.36.0.876.g4bfefc07680
#1 lore raw

Hi Ævar

Quoted 60 lines · Ævar Arnfjörð Bjarmason

Fix a regression in 707d2f2fe86 (CI: use "$runs_on_pool", not "$jobname" to select packages & config, 2021-11-23).

In that commit I changed CC=gcc from CC=gcc-9, but on OSX the "gcc" in $PATH points to clang, we need to use gcc-9 instead. Likewise for the linux-gcc job CC=gcc-8 was changed to the implicit CC=gcc, which would select GCC 9.4.0 instead of GCC 8.4.0.

Furthermore in 25715419bf4 (CI: don't run "make test" twice in one job, 2021-11-23) when the "linux-TEST-vars" job was split off from "linux-gcc" the "cc_package: gcc-8" line was copied along with it.

That wasn't a bug per-se, as that "make test" would have run under GCC 8 before the split into two jobs, but the point of selecting different compiler for these jobs is to get better coverage, and to narrow down any issues with a given compiler to the job that runs it. Since the "linux-TEST-vars" job is already special in other ways (in running with various GIT_TEST_* variables), and we've got the "linux-gcc" job covering gcc-8 let's have it used the default system compiler instead.

---
Range-diff:

1:  94abb826627 < -:  ----------- CI: have osx-gcc use gcc, not clang
-:  ----------- > 1:  d89ad4d5b7c CI: select CC based on CC_PACKAGE (again)

The range-diff is against a relevant patch in v4 of a larger CI series of mine[1]. This fix is independent of that, and on top of master. See
[2] for the original fix and discussion.

As noted in the updated commit message not only the OSX jobs were affected, but also the linux-gcc job. This fixes both, along with a small fix to the related linux-TEST-vars recipe.

1. https://lore.kernel.org/git/[email protected]/
2. https://lore.kernel.org/git/[email protected]/

  .github/workflows/main.yml | 1 -
  ci/lib.sh                  | 2 +-
  2 files changed, 1 insertion(+), 2 deletions(-)
.github/workflows/main.yml +0 -1
index c35200defb9..f12819a00d7 100644
@@ -236,7 +236,6 @@ jobs:
            - jobname: linux-TEST-vars
              cc: gcc
              os: ubuntu
-            cc_package: gcc-8
              pool: ubuntu-latest
            - jobname: osx-clang
              cc: clang
ci/lib.sh +0 -0
index cbc2f8f1caa..44007dcf93b 100755

ci/lib.sh then · line 125

Quoted 5 lines
ci/lib.sh +1 -1
@@ -122,7 +122,7 @@ then
  	test macos != "$CI_OS_NAME" || CI_OS_NAME=osx
  	CI_REPO_SLUG="$GITHUB_REPOSITORY"
  	CI_JOB_ID="$GITHUB_RUN_ID"
-	CC="${CC:-gcc}"
+	CC="${CC:-${CC_PACKAGE:-gcc}}"

CC is set in .github/workflows/main.yaml for the ubuntu and macos jobs so I think they will not fallback to using CC_PACKAGE and therefore not pick up the correct compiler.

Best Wishes

Phillip

Quoted 3 lines
ci/lib.sh +0 -0
  	DONT_SKIP_TAGS=t
 
  	cache_dir="$HOME/none"
#2 lore raw

PATCH v2 1/1 CI: select CC based on CC_PACKAGE (again)

In reply to Ævar Arnfjörð Bjarmason

Fix a regression in 707d2f2fe86 (CI: use "$runs_on_pool", not "$jobname" to select packages & config, 2021-11-23).

In that commit I changed CC=gcc from CC=gcc-9, but on OSX the "gcc" in $PATH points to clang, we need to use gcc-9 instead. Likewise for the linux-gcc job CC=gcc-8 was changed to the implicit CC=gcc, which would select GCC 9.4.0 instead of GCC 8.4.0.

Furthermore in 25715419bf4 (CI: don't run "make test" twice in one job, 2021-11-23) when the "linux-TEST-vars" job was split off from "linux-gcc" the "cc_package: gcc-8" line was copied along with it.

That wasn't a bug per-se, as that "make test" would have run under GCC 8 before the split into two jobs, but the point of selecting different compiler for these jobs is to get better coverage, and to narrow down any issues with a given compiler to the job that runs it. Since the "linux-TEST-vars" job is already special in other ways (in running with various GIT_TEST_* variables), and we've got the "linux-gcc" job covering gcc-8 let's have it used the default system compiler instead.

---

Quoted 3 lines · Phillip Wood

CC is set in .github/workflows/main.yaml for the ubuntu and macos jobs so I think they will not fallback to using CC_PACKAGE and therefore not pick up the correct compiler.

Urgh, sorry. I don't know how I got that turned around, but this version works, and the (currently ongoing) CI run shows that the linux-gcc job selects gcc-8 correctly, instead of gcc:

    https://github.com/avar/git/runs/6116735686?check_suite_focus=true#step:3:22

The osx-gcc job will be likewise fixed, but it's failing in that run due to the unrelated perforce installation issue affecting all OSX jobs (fix here:
https://lore.kernel.org/git/[email protected]/).

Range-diff against v1:
1: d89ad4d5b7c ! 1: 92acf9420a9 CI: select CC based on CC_PACKAGE (again)

    @@ ci/lib.sh: then
      	CI_REPO_SLUG="$GITHUB_REPOSITORY"
      	CI_JOB_ID="$GITHUB_RUN_ID"
     -	CC="${CC:-gcc}"
    -+	CC="${CC:-${CC_PACKAGE:-gcc}}"
    ++	CC="${CC_PACKAGE:-${CC:-gcc}}"
      	DONT_SKIP_TAGS=t
      
      	cache_dir="$HOME/none"
 .github/workflows/main.yml | 1 -
 ci/lib.sh                  | 2 +-
 2 files changed, 1 insertion(+), 2 deletions(-)
.github/workflows/main.yml +0 -1
index c35200defb9..f12819a00d7 100644
@@ -236,7 +236,6 @@ jobs:
           - jobname: linux-TEST-vars
             cc: gcc
             os: ubuntu
-            cc_package: gcc-8
             pool: ubuntu-latest
           - jobname: osx-clang
             cc: clang
ci/lib.sh +1 -1
index cbc2f8f1caa..86e37da9bc5 100755
@@ -122,7 +122,7 @@ then
 	test macos != "$CI_OS_NAME" || CI_OS_NAME=osx
 	CI_REPO_SLUG="$GITHUB_REPOSITORY"
 	CI_JOB_ID="$GITHUB_RUN_ID"
-	CC="${CC:-gcc}"
+	CC="${CC_PACKAGE:-${CC:-gcc}}"
 	DONT_SKIP_TAGS=t
 
 	cache_dir="$HOME/none"
Signature · 1 line
2.36.0.879.g3659959fcca
#3 lore raw
Quoted 2 lines · Ævar Arnfjörð Bjarmason

to avoid confusing the attribution scripts better use my full name:
"Carlo Marcelo Arenas Belón"

ci/lib.sh then · line 125

Quoted 10 lines
ci/lib.sh +1 -1
index cbc2f8f1caa..86e37da9bc5 100755
@@ -122,7 +122,7 @@ then
        test macos != "$CI_OS_NAME" || CI_OS_NAME=osx
        CI_REPO_SLUG="$GITHUB_REPOSITORY"
        CI_JOB_ID="$GITHUB_RUN_ID"
-       CC="${CC:-gcc}"
+       CC="${CC_PACKAGE:-${CC:-gcc}}"

minor nitpick, but most likely still relevant considering your other "bashism" fixes.
the POSIX syntax doesn't use ":" (documented in CodingGuidelines)

not sure if the compounded assignment is valid in POSIX, but at least the following worked with a recent NetBSD 8 sh :

  CC="${CC_PACKAGE-${CC-gcc}}"

Carlo

#4 lore raw
Quoted 6 lines · Carlo Arenas

ci/lib.sh then · line 125

Quoted 2 lines
ci/lib.sh +1 -1
-       CC="${CC:-gcc}"
+       CC="${CC_PACKAGE:-${CC:-gcc}}"

minor nitpick, but most likely still relevant considering your other "bashism" fixes.
the POSIX syntax doesn't use ":" (documented in CodingGuidelines)

You are reading the guideline wrong, I am afraid.

This is not a substring substitution, ${foo:0:3} picking three characters from the beginning of foo.

It is "If CC_PACKAGE is UNSET OR SET TO EMPTY, then use the fallback value that is ${CC:-gcc}", which is a bog standard bourne shell that predates POSIX. The colon before "-" is what adds "OR SET TO EMPTY" part; without the colon, i.e. "${CC_PACKAGE-${CC-gcc}}", then you get the same without "OR SET TO EMPTY" part.

And we do use that form.

#5 lore raw
Quoted 11 lines · Junio C Hamano
Quoted 6 lines · Carlo Arenas

ci/lib.sh then · line 125

Quoted 2 lines
ci/lib.sh +1 -1
-       CC="${CC:-gcc}"
+       CC="${CC_PACKAGE:-${CC:-gcc}}"

minor nitpick, but most likely still relevant considering your other "bashism" fixes.
the POSIX syntax doesn't use ":" (documented in CodingGuidelines)

You are reading the guideline wrong, I am afraid.

Indeed; and I realized it as I was trying to answer my own question about the compounded replacement being supported.

Wondering if something more explicit might be easier to understand anyway like :

if test -n "$CC_PACKAGE"
then

     CC="$CC_PACKAGE"

fi

Carlo

#6 lore raw
Quoted 7 lines · Carlo Arenas
Quoted 6 lines · Junio C Hamano
Quoted 3 lines · Carlo Arenas
Quoted 2 lines
ci/lib.sh +1 -1
-       CC="${CC:-gcc}"
+       CC="${CC_PACKAGE:-${CC:-gcc}}"

...

Quoted 8 lines

Wondering if something more explicit might be easier to understand anyway like :

if test -n "$CC_PACKAGE"
then

     CC="$CC_PACKAGE"

fi

Carlo

The parameter substitution is commonly used part of the shell language, and even if developers, when starting to work on parts of the system that are written in shell, are initially unfamiliar with such language constructs, nobody will stay a newbie forever.

To them, idioms like ${A:-${B:-C}} would quickly become as explicit as a long-form written as if/then/elif/fi. Surely it takes practice to be familiar with the idiom, but idioms help those who know them to be more efficient and effetive, so, no, I do not think it is a good idea to aim for "easier to understand by newbies".

Thanks.

#7 lore raw
In reply to Ævar Arnfjörð Bjarmason
Quoted 7 lines · Ævar Arnfjörð Bjarmason

Fix a regression in 707d2f2fe86 (CI: use "$runs_on_pool", not "$jobname" to select packages & config, 2021-11-23).

In that commit I changed CC=gcc from CC=gcc-9, but on OSX the "gcc" in $PATH points to clang, we need to use gcc-9 instead. Likewise for the linux-gcc job CC=gcc-8 was changed to the implicit CC=gcc, which would select GCC 9.4.0 instead of GCC 8.4.0.

Thanks for diagnosing how things were broken.

Quoted 23 lines · Phillip Wood
Quoted 4 lines

CC is set in .github/workflows/main.yaml for the ubuntu and macos jobs so I think they will not fallback to using CC_PACKAGE and therefore not pick up the correct compiler.

...

.github/workflows/main.yml +0 -1
index c35200defb9..f12819a00d7 100644
@@ -236,7 +236,6 @@ jobs:
           - jobname: linux-TEST-vars
             cc: gcc
             os: ubuntu
-            cc_package: gcc-8
             pool: ubuntu-latest
           - jobname: osx-clang
             cc: clang
ci/lib.sh +0 -0
index cbc2f8f1caa..86e37da9bc5 100755

ci/lib.sh then · line 128

Quoted 8 lines
ci/lib.sh +1 -1
@@ -122,7 +122,7 @@ then
 	test macos != "$CI_OS_NAME" || CI_OS_NAME=osx
 	CI_REPO_SLUG="$GITHUB_REPOSITORY"
 	CI_JOB_ID="$GITHUB_RUN_ID"
-	CC="${CC:-gcc}"
+	CC="${CC_PACKAGE:-${CC:-gcc}}"
 	DONT_SKIP_TAGS=t
 
 	cache_dir="$HOME/none"

OK, so we favor CC_PACKAGE (from the matrix.vector.cc_package) if set, and then cc (again, from the matrix.vector.cc) if set, and then finally use "gcc" as a fallback. In the osx-gcc job, cc_package is set to gcc-9 while in the osx-clang, cc is gcc that confusingly calls for clang. That sounds like it would do the right thing for two macs.

For other jobs with different settings for cc and cc_package, does this have any effect? I do not think I saw any mention in the proposed log message.

vector.cc vector.cc_package old new
linux-clang clang - clang clang
linux-sha256 clang - clang clang
linux-gcc gcc gcc-8 gcc gcc-8
osx-clang clang - clang clang
osx-gcc gcc gcc-9 clang gcc-9
linux-gcc-default gcc - gcc gcc

So, linux-gcc job used to use whichever "gcc" the platform gave us, but now it explicitly asks for gcc-8, which may or may not be different from what linux-gcc-default uses, and there is no other difference by this change. We may get a better test coverage (if the default gcc is not gcc-8) or no improvement (if the default is gcc-8), so it is a strict improvement worth recording as an intended side effect in the proposed log message to help future developers.

Other than that, looks good to me.

#8 lore raw
Quoted 42 lines · Junio C Hamano
Quoted 7 lines · Ævar Arnfjörð Bjarmason

Fix a regression in 707d2f2fe86 (CI: use "$runs_on_pool", not "$jobname" to select packages & config, 2021-11-23).

In that commit I changed CC=gcc from CC=gcc-9, but on OSX the "gcc" in $PATH points to clang, we need to use gcc-9 instead. Likewise for the linux-gcc job CC=gcc-8 was changed to the implicit CC=gcc, which would select GCC 9.4.0 instead of GCC 8.4.0.

Thanks for diagnosing how things were broken.

Quoted 30 lines · Phillip Wood
Quoted 4 lines

CC is set in .github/workflows/main.yaml for the ubuntu and macos jobs so I think they will not fallback to using CC_PACKAGE and therefore not pick up the correct compiler.

...

.github/workflows/main.yml +0 -1
index c35200defb9..f12819a00d7 100644
@@ -236,7 +236,6 @@ jobs:
           - jobname: linux-TEST-vars
             cc: gcc
             os: ubuntu
-            cc_package: gcc-8
             pool: ubuntu-latest
           - jobname: osx-clang
             cc: clang
ci/lib.sh +1 -1
index cbc2f8f1caa..86e37da9bc5 100755
@@ -122,7 +122,7 @@ then
 	test macos != "$CI_OS_NAME" || CI_OS_NAME=osx
 	CI_REPO_SLUG="$GITHUB_REPOSITORY"
 	CI_JOB_ID="$GITHUB_RUN_ID"
-	CC="${CC:-gcc}"
+	CC="${CC_PACKAGE:-${CC:-gcc}}"
 	DONT_SKIP_TAGS=t
 
Quoted 8 lines

ci/lib.sh then · line 128

Quoted 1 line
ci/lib.sh +0 -0
 	cache_dir="$HOME/none"

OK, so we favor CC_PACKAGE (from the matrix.vector.cc_package) if set, and then cc (again, from the matrix.vector.cc) if set, and then finally use "gcc" as a fallback. In the osx-gcc job, cc_package is set to gcc-9 while in the osx-clang, cc is gcc that confusingly calls for clang. That sounds like it would do the right thing for two macs.

Yes.

Quoted 20 lines

For other jobs with different settings for cc and cc_package, does this have any effect? I do not think I saw any mention in the proposed log message.

vector.cc vector.cc_package old new
linux-clang clang - clang clang
linux-sha256 clang - clang clang
linux-gcc gcc gcc-8 gcc gcc-8
osx-clang clang - clang clang
osx-gcc gcc gcc-9 clang gcc-9
linux-gcc-default gcc - gcc gcc

So, linux-gcc job used to use whichever "gcc" the platform gave us, but now it explicitly asks for gcc-8, which may or may not be different from what linux-gcc-default uses, and there is no other difference by this change. We may get a better test coverage (if the default gcc is not gcc-8) or no improvement (if the default is gcc-8), so it is a strict improvement worth recording as an intended side effect in the proposed log message to help future developers.

Quoted 1 line

Other than that, looks good to me.

I'm happy to rephrase it however you'd like, but I'm a bit confused by the "saw any mention in the proposed log message". I'm fairly sure paragraph 2 onwards covers this, i.e. how linux-gcc's behavior is changed (as it also regressed).

What I suppose is left undiscussed is that jobs that don't define CC_PACKAGE at all won't be impacted, is that what you wanted to be explicitly mentioned?

#9 lore raw
Quoted 4 lines · Ævar Arnfjörð Bjarmason

I'm happy to rephrase it however you'd like, but I'm a bit confused by the "saw any mention in the proposed log message". I'm fairly sure paragraph 2 onwards covers this, i.e. how linux-gcc's behavior is changed (as it also regressed).

Yeah, true, """Likewise for the linux-gcc job CC=gcc-8 was changed to the implicit CC=gcc, which would select GCC 9.4.0 instead of GCC 8.4.0.""" that you wrote is exactly about it.

I was confused because immediately after that you said it was not a bug, so I dismissed it as not part or the real issue. Perhaps striking that "not a bug" part may make it less confusing? I dunno.

Quoted 3 lines

What I suppose is left undiscussed is that jobs that don't define CC_PACKAGE at all won't be impacted, is that what you wanted to be explicitly mentioned?

Yes, I agree that is a good idea, too. Or you can steal the "before and after" table I gave you in the message you are responding to, if you think it helps.

Thanks.

#10 lore raw
Quoted 7 lines · Junio C Hamano
Quoted 4 lines · Ævar Arnfjörð Bjarmason

I'm happy to rephrase it however you'd like, but I'm a bit confused by the "saw any mention in the proposed log message". I'm fairly sure paragraph 2 onwards covers this, i.e. how linux-gcc's behavior is changed (as it also regressed).

Quoted 7 lines

Yeah, true, """Likewise for the linux-gcc job CC=gcc-8 was changed to the implicit CC=gcc, which would select GCC 9.4.0 instead of GCC 8.4.0.""" that you wrote is exactly about it.

I was confused because immediately after that you said it was not a bug, so I dismissed it as not part or the real issue. Perhaps striking that "not a bug" part may make it less confusing? I dunno.

I'll rephrase it, but I probably won't get to it tonight, sorry.

What I was trying to get at is that the linux-TEST-vars job was affected, and *does* have a CC_PACKAGE defined, but in that case (and I was the one who split it out) the point was never to have it use its own compiler. For any of these platform/compiler combinations it's enough if we cover that platform/compiler verison once.

So the right thing to do with linux-TEST-vars is to just have it use "vanilla", because it's special in other ways. So if and when it fails we don't want to worry about untangling the two variables. I.e. is it the special-versioned compiler, or is it (and it almost always would be) the GIT_TEST_* variables. In that case we should only need to worry about the latter.

Quoted 7 lines
Quoted 3 lines

What I suppose is left undiscussed is that jobs that don't define CC_PACKAGE at all won't be impacted, is that what you wanted to be explicitly mentioned?

Yes, I agree that is a good idea, too. Or you can steal the "before and after" table I gave you in the message you are responding to, if you think it helps.

*nod*

#11 lore raw
In reply to Junio C Hamano
Quoted 7 lines · Junio C Hamano
Quoted 4 lines · Ævar Arnfjörð Bjarmason

I'm happy to rephrase it however you'd like, but I'm a bit confused by the "saw any mention in the proposed log message". I'm fairly sure paragraph 2 onwards covers this, i.e. how linux-gcc's behavior is changed (as it also regressed).

Quoted 7 lines

Yeah, true, """Likewise for the linux-gcc job CC=gcc-8 was changed to the implicit CC=gcc, which would select GCC 9.4.0 instead of GCC 8.4.0.""" that you wrote is exactly about it.

I was confused because immediately after that you said it was not a bug, so I dismissed it as not part or the real issue. Perhaps striking that "not a bug" part may make it less confusing? I dunno.

Also, the description above merely says we use gcc-9 instead of gcc-8, without saying if that is a good change or not, which is another thing that needs to be said. In the larger context, if we aim to try different compilers, we should know which other ones we are using, and if linux-gcc job is meant to test gcc-8 (because we know there is another job that tests gcc-9), the change was a bug. But because you said it was not a bug, it made it unclear if that is merely an observation or it was pointing out our earlier mistake that should be corrected.

#12 lore raw

PATCH v3 1/1 CI: select CC based on CC_PACKAGE (again)

In reply to Ævar Arnfjörð Bjarmason

Fix a regression in 707d2f2fe86 (CI: use "$runs_on_pool", not "$jobname" to select packages & config, 2021-11-23).

In that commit I changed CC=gcc from CC=gcc-9, but on OSX the "gcc" in $PATH points to clang, we need to use gcc-9 instead. Likewise for the linux-gcc job CC=gcc-8 was changed to the implicit CC=gcc, which would select GCC 9.4.0 instead of GCC 8.4.0.

Furthermore in 25715419bf4 (CI: don't run "make test" twice in one job, 2021-11-23) when the "linux-TEST-vars" job was split off from "linux-gcc" the "cc_package: gcc-8" line was copied along with it, so its "cc_package" line wasn't working as intended either.

As a table, this is what's changed by this commit, i.e. it only affects the linux-gcc, linux-TEST-vars and osx-gcc jobs:

	|-------------------+-----------+-------------------+-------+-------|
	| jobname           | vector.cc | vector.cc_package | old   | new   |
	|-------------------+-----------+-------------------+-------+-------|
	| linux-clang       | clang     | -                 | clang | clang |
	| linux-sha256      | clang     | -                 | clang | clang |
	| linux-gcc         | gcc       | gcc-8             | gcc   | gcc-8 |
	| osx-clang         | clang     | -                 | clang | clang |
	| osx-gcc           | gcc       | gcc-9             | clang | gcc-9 |
	| linux-gcc-default | gcc       | -                 | gcc   | gcc   |
	| linux-TEST-vars   | gcc       | gcc-8             | gcc   | gcc-8 |
	|-------------------+-----------+-------------------+-------+-------|

---

I just dropped the linux-TEST-vars change from the v2 in lieu of trying to get the wording in the commit message right.

Range-diff against v2:
1: 92acf9420a9 ! 1: 8b3444ecc87 CI: select CC based on CC_PACKAGE (again)
@@ Commit message
Furthermore in 25715419bf4 (CI: don't run "make test" twice in one job, 2021-11-23) when the "linux-TEST-vars" job was split off from "linux-gcc" the "cc_package: gcc-8" line was copied along with
- it.
-
- That wasn't a bug per-se, as that "make test" would have run under GCC
- 8 before the split into two jobs, but the point of selecting different
- compiler for these jobs is to get better coverage, and to narrow down
- any issues with a given compiler to the job that runs it. Since the
- "linux-TEST-vars" job is already special in other ways (in running
- with various GIT_TEST_* variables), and we've got the "linux-gcc" job
- covering gcc-8 let's have it used the default system compiler instead.
+ it, so its "cc_package" line wasn't working as intended either.
+
+ As a table, this is what's changed by this commit, i.e. it only
+ affects the linux-gcc, linux-TEST-vars and osx-gcc jobs:
+
+ |-------------------+-----------+-------------------+-------+-------|
+ | jobname | vector.cc | vector.cc_package | old | new |
+ |-------------------+-----------+-------------------+-------+-------|
+ | linux-clang | clang | - | clang | clang |
+ | linux-sha256 | clang | - | clang | clang |
+ | linux-gcc | gcc | gcc-8 | gcc | gcc-8 |
+ | osx-clang | clang | - | clang | clang |
+ | osx-gcc | gcc | gcc-9 | clang | gcc-9 |
+ | linux-gcc-default | gcc | - | gcc | gcc |
+ | linux-TEST-vars | gcc | gcc-8 | gcc | gcc-8 |
+ |-------------------+-----------+-------------------+-------+-------|

- ## .github/workflows/main.yml ##
-@@ .github/workflows/main.yml: jobs:
- - jobname: linux-TEST-vars
- cc: gcc
- os: ubuntu
-- cc_package: gcc-8
- pool: ubuntu-latest
- - jobname: osx-clang
- cc: clang
-
## ci/lib.sh ##
@@ ci/lib.sh: then
test macos != "$CI_OS_NAME" || CI_OS_NAME=osx

 ci/lib.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
ci/lib.sh +1 -1
index cbc2f8f1caa..86e37da9bc5 100755
@@ -122,7 +122,7 @@ then
 	test macos != "$CI_OS_NAME" || CI_OS_NAME=osx
 	CI_REPO_SLUG="$GITHUB_REPOSITORY"
 	CI_JOB_ID="$GITHUB_RUN_ID"
-	CC="${CC:-gcc}"
+	CC="${CC_PACKAGE:-${CC:-gcc}}"
 	DONT_SKIP_TAGS=t
 
 	cache_dir="$HOME/none"
Signature · 1 line
2.36.0.879.g56a83971f3f
#13 lore raw

Hi Ævar

Quoted 19 lines · On 22 April 2022 at 10:20 Ævar Arnfjörð Bjarmason <[email protected]> wrote:

Fix a regression in 707d2f2fe86 (CI: use "$runs_on_pool", not "$jobname" to select packages & config, 2021-11-23).

In that commit I changed CC=gcc from CC=gcc-9, but on OSX the "gcc" in $PATH points to clang, we need to use gcc-9 instead. Likewise for the linux-gcc job CC=gcc-8 was changed to the implicit CC=gcc, which would select GCC 9.4.0 instead of GCC 8.4.0.

Furthermore in 25715419bf4 (CI: don't run "make test" twice in one job, 2021-11-23) when the "linux-TEST-vars" job was split off from "linux-gcc" the "cc_package: gcc-8" line was copied along with it, so its "cc_package" line wasn't working as intended either.

As a table, this is what's changed by this commit, i.e. it only affects the linux-gcc, linux-TEST-vars and osx-gcc jobs:

Quoted 11 lines
	|-------------------+-----------+-------------------+-------+-------|
	| jobname           | vector.cc | vector.cc_package | old   | new   |
	|-------------------+-----------+-------------------+-------+-------|
	| linux-clang       | clang     | -                 | clang | clang |
	| linux-sha256      | clang     | -                 | clang | clang |
	| linux-gcc         | gcc       | gcc-8             | gcc   | gcc-8 |
	| osx-clang         | clang     | -                 | clang | clang |
	| osx-gcc           | gcc       | gcc-9             | clang | gcc-9 |
	| linux-gcc-default | gcc       | -                 | gcc   | gcc   |
	| linux-TEST-vars   | gcc       | gcc-8             | gcc   | gcc-8 |
	|-------------------+-----------+-------------------+-------+-------|

Having this table is helpful. I do find it confusing that we're still setting CC in the main.yaml and then overriding it in lib.sh. I think it would make things clearer if we got find of the cc settings in the job matrix.

Best Wishes

Phillip

Quoted 75 lines

---

I just dropped the linux-TEST-vars change from the v2 in lieu of trying to get the wording in the commit message right.

Range-diff against v2:
1: 92acf9420a9 ! 1: 8b3444ecc87 CI: select CC based on CC_PACKAGE (again)
@@ Commit message
Furthermore in 25715419bf4 (CI: don't run "make test" twice in one job, 2021-11-23) when the "linux-TEST-vars" job was split off from "linux-gcc" the "cc_package: gcc-8" line was copied along with
- it.
-
- That wasn't a bug per-se, as that "make test" would have run under GCC
- 8 before the split into two jobs, but the point of selecting different
- compiler for these jobs is to get better coverage, and to narrow down
- any issues with a given compiler to the job that runs it. Since the
- "linux-TEST-vars" job is already special in other ways (in running
- with various GIT_TEST_* variables), and we've got the "linux-gcc" job
- covering gcc-8 let's have it used the default system compiler instead.
+ it, so its "cc_package" line wasn't working as intended either.
+
+ As a table, this is what's changed by this commit, i.e. it only
+ affects the linux-gcc, linux-TEST-vars and osx-gcc jobs:
+
+ |-------------------+-----------+-------------------+-------+-------|
+ | jobname | vector.cc | vector.cc_package | old | new |
+ |-------------------+-----------+-------------------+-------+-------|
+ | linux-clang | clang | - | clang | clang |
+ | linux-sha256 | clang | - | clang | clang |
+ | linux-gcc | gcc | gcc-8 | gcc | gcc-8 |
+ | osx-clang | clang | - | clang | clang |
+ | osx-gcc | gcc | gcc-9 | clang | gcc-9 |
+ | linux-gcc-default | gcc | - | gcc | gcc |
+ | linux-TEST-vars | gcc | gcc-8 | gcc | gcc-8 |
+ |-------------------+-----------+-------------------+-------+-------|

- ## .github/workflows/main.yml ##
-@@ .github/workflows/main.yml: jobs:
- - jobname: linux-TEST-vars
- cc: gcc
- os: ubuntu
-- cc_package: gcc-8
- pool: ubuntu-latest
- - jobname: osx-clang
- cc: clang
-
## ci/lib.sh ##
@@ ci/lib.sh: then
test macos != "$CI_OS_NAME" || CI_OS_NAME=osx

 ci/lib.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
ci/lib.sh +1 -1
index cbc2f8f1caa..86e37da9bc5 100755
@@ -122,7 +122,7 @@ then
 	test macos != "$CI_OS_NAME" || CI_OS_NAME=osx
 	CI_REPO_SLUG="$GITHUB_REPOSITORY"
 	CI_JOB_ID="$GITHUB_RUN_ID"
-	CC="${CC:-gcc}"
+	CC="${CC_PACKAGE:-${CC:-gcc}}"
 	DONT_SKIP_TAGS=t
 
 	cache_dir="$HOME/none"
Signature · 1 line
2.36.0.879.g56a83971f3f
#14 lore raw
Quoted 33 lines · Phillip Wood

Hi Ævar

Quoted 19 lines · On 22 April 2022 at 10:20 Ævar Arnfjörð Bjarmason <[email protected]> wrote:

Fix a regression in 707d2f2fe86 (CI: use "$runs_on_pool", not "$jobname" to select packages & config, 2021-11-23).

In that commit I changed CC=gcc from CC=gcc-9, but on OSX the "gcc" in $PATH points to clang, we need to use gcc-9 instead. Likewise for the linux-gcc job CC=gcc-8 was changed to the implicit CC=gcc, which would select GCC 9.4.0 instead of GCC 8.4.0.

Furthermore in 25715419bf4 (CI: don't run "make test" twice in one job, 2021-11-23) when the "linux-TEST-vars" job was split off from "linux-gcc" the "cc_package: gcc-8" line was copied along with it, so its "cc_package" line wasn't working as intended either.

As a table, this is what's changed by this commit, i.e. it only affects the linux-gcc, linux-TEST-vars and osx-gcc jobs:

Quoted 11 lines
	|-------------------+-----------+-------------------+-------+-------|
	| jobname           | vector.cc | vector.cc_package | old   | new   |
	|-------------------+-----------+-------------------+-------+-------|
	| linux-clang       | clang     | -                 | clang | clang |
	| linux-sha256      | clang     | -                 | clang | clang |
	| linux-gcc         | gcc       | gcc-8             | gcc   | gcc-8 |
	| osx-clang         | clang     | -                 | clang | clang |
	| osx-gcc           | gcc       | gcc-9             | clang | gcc-9 |
	| linux-gcc-default | gcc       | -                 | gcc   | gcc   |
	| linux-TEST-vars   | gcc       | gcc-8             | gcc   | gcc-8 |
	|-------------------+-----------+-------------------+-------+-------|
Quoted 3 lines

Having this table is helpful. I do find it confusing that we're still setting CC in the main.yaml and then overriding it in lib.sh. I think it would make things clearer if we got find of the cc settings in the job matrix.

Yes, this stand-alone patch is just a band-aid.

Do you like the state at the tip of my larger CI series? It's only set in ci/lib.sh there, or in the case of the package we need the installation script is tasked with it now.

#15 lore raw
In reply to Ævar Arnfjörð Bjarmason
Quoted 28 lines · Ævar Arnfjörð Bjarmason

Fix a regression in 707d2f2fe86 (CI: use "$runs_on_pool", not "$jobname" to select packages & config, 2021-11-23).

In that commit I changed CC=gcc from CC=gcc-9, but on OSX the "gcc" in $PATH points to clang, we need to use gcc-9 instead. Likewise for the linux-gcc job CC=gcc-8 was changed to the implicit CC=gcc, which would select GCC 9.4.0 instead of GCC 8.4.0.

Furthermore in 25715419bf4 (CI: don't run "make test" twice in one job, 2021-11-23) when the "linux-TEST-vars" job was split off from "linux-gcc" the "cc_package: gcc-8" line was copied along with it, so its "cc_package" line wasn't working as intended either.

As a table, this is what's changed by this commit, i.e. it only affects the linux-gcc, linux-TEST-vars and osx-gcc jobs:

	|-------------------+-----------+-------------------+-------+-------|
	| jobname           | vector.cc | vector.cc_package | old   | new   |
	|-------------------+-----------+-------------------+-------+-------|
	| linux-clang       | clang     | -                 | clang | clang |
	| linux-sha256      | clang     | -                 | clang | clang |
	| linux-gcc         | gcc       | gcc-8             | gcc   | gcc-8 |
	| osx-clang         | clang     | -                 | clang | clang |
	| osx-gcc           | gcc       | gcc-9             | clang | gcc-9 |
	| linux-gcc-default | gcc       | -                 | gcc   | gcc   |
	| linux-TEST-vars   | gcc       | gcc-8             | gcc   | gcc-8 |
	|-------------------+-----------+-------------------+-------+-------|
Quoted 6 lines

---

I just dropped the linux-TEST-vars change from the v2 in lieu of trying to get the wording in the commit message right.

OK. So linux-gcc and linux-gcc-default jobs use different compilers (if gcc and gcc-8 are different), osx-gcc uses gcc-9 instead of clang.

Thanks. Will queue.

Quoted 13 lines
ci/lib.sh +1 -1
index cbc2f8f1caa..86e37da9bc5 100755
@@ -122,7 +122,7 @@ then
 	test macos != "$CI_OS_NAME" || CI_OS_NAME=osx
 	CI_REPO_SLUG="$GITHUB_REPOSITORY"
 	CI_JOB_ID="$GITHUB_RUN_ID"
-	CC="${CC:-gcc}"
+	CC="${CC_PACKAGE:-${CC:-gcc}}"
 	DONT_SKIP_TAGS=t
 
 	cache_dir="$HOME/none"
#16 lore raw
In reply to Ævar Arnfjörð Bjarmason
Quoted 28 lines · Ævar Arnfjörð Bjarmason

Fix a regression in 707d2f2fe86 (CI: use "$runs_on_pool", not "$jobname" to select packages & config, 2021-11-23).

In that commit I changed CC=gcc from CC=gcc-9, but on OSX the "gcc" in $PATH points to clang, we need to use gcc-9 instead. Likewise for the linux-gcc job CC=gcc-8 was changed to the implicit CC=gcc, which would select GCC 9.4.0 instead of GCC 8.4.0.

Furthermore in 25715419bf4 (CI: don't run "make test" twice in one job, 2021-11-23) when the "linux-TEST-vars" job was split off from "linux-gcc" the "cc_package: gcc-8" line was copied along with it, so its "cc_package" line wasn't working as intended either.

As a table, this is what's changed by this commit, i.e. it only affects the linux-gcc, linux-TEST-vars and osx-gcc jobs:

	|-------------------+-----------+-------------------+-------+-------|
	| jobname           | vector.cc | vector.cc_package | old   | new   |
	|-------------------+-----------+-------------------+-------+-------|
	| linux-clang       | clang     | -                 | clang | clang |
	| linux-sha256      | clang     | -                 | clang | clang |
	| linux-gcc         | gcc       | gcc-8             | gcc   | gcc-8 |
	| osx-clang         | clang     | -                 | clang | clang |
	| osx-gcc           | gcc       | gcc-9             | clang | gcc-9 |
	| linux-gcc-default | gcc       | -                 | gcc   | gcc   |
	| linux-TEST-vars   | gcc       | gcc-8             | gcc   | gcc-8 |
	|-------------------+-----------+-------------------+-------+-------|
Quoted 6 lines

---

I just dropped the linux-TEST-vars change from the v2 in lieu of trying to get the wording in the commit message right.

I plan to fast-track this; the revamp of ci/lib.sh in the other huge series would probably need to be rebased on top, or independently fix it the same way.

In the meantime, here is what I'll throwing into today's 'seen'.

ci/lib.sh +0 -0
diff --cc ci/lib.sh
index 86e37da9bc,80e89f89b7..0000000000
@@@ -195,6 -224,79 +224,79 @@@ ubuntu-latest
  esac
  
  case "$jobname" in
+ windows-build)
+ 	setenv --build NO_PERL NoThanks
+ 	setenv --build ARTIFACTS_DIRECTORY artifacts
+ 	;;
+ windows-test)
+ 	setenv --test MAKEFLAGS "$COMMON_MAKEFLAGS"
+ 	;;
+ vs-build)
+ 	setenv --build DEVELOPER $DEVELOPER
+ 	setenv --build SKIP_DASHED_BUILT_INS $SKIP_DASHED_BUILT_INS
+ 
+ 	setenv --build NO_PERL NoThanks
+ 	setenv --build NO_GETTEXT NoThanks
+ 	setenv --build ARTIFACTS_DIRECTORY artifacts
+ 	setenv --build INCLUDE_DLLS_IN_ARTIFACTS YesPlease
+ 	setenv --build MSVC YesPlease
+ 
+ 	setenv --build GIT_CONFIG_COUNT 2
+ 	setenv --build GIT_CONFIG_KEY_0 user.name
+ 	setenv --build GIT_CONFIG_VALUE_0 CI
+ 	setenv --build GIT_CONFIG_KEY_1 user.emailname
+ 	setenv --build GIT_CONFIG_VALUE_1 ci@git
+ 	setenv --build GIT_CONFIG_VALUE_1 ci@git
+ 	;;
+ vs-test)
+ 	setenv --test NO_SVN_TESTS YesPlease
+ 	setenv --test MAKEFLAGS "$COMMON_MAKEFLAGS"
+ 	;;
+ linux-gcc)
 -	CC=gcc
++	CC=gcc-8
+ 	setenv --test GIT_TEST_DEFAULT_INITIAL_BRANCH_NAME main
+ 	;;
+ linux-gcc-default)
+ 	CC=gcc
+ 	;;
+ linux-TEST-vars)
 -	CC=gcc
++	CC=gcc-8
+ 	setenv --test GIT_TEST_SPLIT_INDEX yes
+ 	setenv --test GIT_TEST_MERGE_ALGORITHM recursive
+ 	setenv --test GIT_TEST_FULL_IN_PACK_ARRAY true
+ 	setenv --test GIT_TEST_OE_SIZE 10
+ 	setenv --test GIT_TEST_OE_DELTA_SIZE 5
+ 	setenv --test GIT_TEST_COMMIT_GRAPH 1
+ 	setenv --test GIT_TEST_COMMIT_GRAPH_CHANGED_PATHS 1
+ 	setenv --test GIT_TEST_MULTI_PACK_INDEX 1
+ 	setenv --test GIT_TEST_MULTI_PACK_INDEX_WRITE_BITMAP 1
+ 	setenv --test GIT_TEST_ADD_I_USE_BUILTIN 1
+ 	setenv --test GIT_TEST_DEFAULT_INITIAL_BRANCH_NAME master
+ 	setenv --test GIT_TEST_WRITE_REV_INDEX 1
+ 	setenv --test GIT_TEST_CHECKOUT_WORKERS 2
+ 	;;
+ osx-gcc)
+ 	MAKEFLAGS="$MAKEFLAGS PYTHON_PATH=$(which python3)"
 -	CC=gcc
++	CC=gcc-9
+ 	;;
+ osx-clang)
+ 	MAKEFLAGS="$MAKEFLAGS PYTHON_PATH=$(which python2)"
+ 	CC=clang
+ 	;;
+ linux-clang)
+ 	CC=clang
+ 	setenv --test GIT_TEST_DEFAULT_HASH sha1
+ 	;;
+ linux-sha256)
+ 	CC=clang
+ 	setenv --test GIT_TEST_DEFAULT_HASH sha256
+ 	;;
+ pedantic)
+ 	CC=gcc
+ 	# Don't run the tests; we only care about whether Git can be
+ 	# built.
+ 	setenv --build DEVOPTS pedantic
+ 	;;
  linux32)
  	CC=gcc
  	;;
#17 lore raw
Quoted 120 lines · Junio C Hamano
Quoted 28 lines · Ævar Arnfjörð Bjarmason

Fix a regression in 707d2f2fe86 (CI: use "$runs_on_pool", not "$jobname" to select packages & config, 2021-11-23).

In that commit I changed CC=gcc from CC=gcc-9, but on OSX the "gcc" in $PATH points to clang, we need to use gcc-9 instead. Likewise for the linux-gcc job CC=gcc-8 was changed to the implicit CC=gcc, which would select GCC 9.4.0 instead of GCC 8.4.0.

Furthermore in 25715419bf4 (CI: don't run "make test" twice in one job, 2021-11-23) when the "linux-TEST-vars" job was split off from "linux-gcc" the "cc_package: gcc-8" line was copied along with it, so its "cc_package" line wasn't working as intended either.

As a table, this is what's changed by this commit, i.e. it only affects the linux-gcc, linux-TEST-vars and osx-gcc jobs:

	|-------------------+-----------+-------------------+-------+-------|
	| jobname           | vector.cc | vector.cc_package | old   | new   |
	|-------------------+-----------+-------------------+-------+-------|
	| linux-clang       | clang     | -                 | clang | clang |
	| linux-sha256      | clang     | -                 | clang | clang |
	| linux-gcc         | gcc       | gcc-8             | gcc   | gcc-8 |
	| osx-clang         | clang     | -                 | clang | clang |
	| osx-gcc           | gcc       | gcc-9             | clang | gcc-9 |
	| linux-gcc-default | gcc       | -                 | gcc   | gcc   |
	| linux-TEST-vars   | gcc       | gcc-8             | gcc   | gcc-8 |
	|-------------------+-----------+-------------------+-------+-------|
Quoted 6 lines

---

I just dropped the linux-TEST-vars change from the v2 in lieu of trying to get the wording in the commit message right.

I plan to fast-track this; the revamp of ci/lib.sh in the other huge series would probably need to be rebased on top, or independently fix it the same way.

In the meantime, here is what I'll throwing into today's 'seen'.

ci/lib.sh +66 -0
diff --cc ci/lib.sh
index 86e37da9bc,80e89f89b7..0000000000
@@@ -195,6 -224,79 +224,79 @@@ ubuntu-latest
  esac
 
  case "$jobname" in
+ windows-build)
+ 	setenv --build NO_PERL NoThanks
+ 	setenv --build ARTIFACTS_DIRECTORY artifacts
+ 	;;
+ windows-test)
+ 	setenv --test MAKEFLAGS "$COMMON_MAKEFLAGS"
+ 	;;
+ vs-build)
+ 	setenv --build DEVELOPER $DEVELOPER
+ 	setenv --build SKIP_DASHED_BUILT_INS $SKIP_DASHED_BUILT_INS
+ 
+ 	setenv --build NO_PERL NoThanks
+ 	setenv --build NO_GETTEXT NoThanks
+ 	setenv --build ARTIFACTS_DIRECTORY artifacts
+ 	setenv --build INCLUDE_DLLS_IN_ARTIFACTS YesPlease
+ 	setenv --build MSVC YesPlease
+ 
+ 	setenv --build GIT_CONFIG_COUNT 2
+ 	setenv --build GIT_CONFIG_KEY_0 user.name
+ 	setenv --build GIT_CONFIG_VALUE_0 CI
+ 	setenv --build GIT_CONFIG_KEY_1 user.emailname
+ 	setenv --build GIT_CONFIG_VALUE_1 ci@git
+ 	setenv --build GIT_CONFIG_VALUE_1 ci@git
+ 	;;
+ vs-test)
+ 	setenv --test NO_SVN_TESTS YesPlease
+ 	setenv --test MAKEFLAGS "$COMMON_MAKEFLAGS"
+ 	;;
+ linux-gcc)
 -	CC=gcc
++	CC=gcc-8
+ 	setenv --test GIT_TEST_DEFAULT_INITIAL_BRANCH_NAME main
+ 	;;
+ linux-gcc-default)
+ 	CC=gcc
+ 	;;
+ linux-TEST-vars)
 -	CC=gcc
++	CC=gcc-8
+ 	setenv --test GIT_TEST_SPLIT_INDEX yes
+ 	setenv --test GIT_TEST_MERGE_ALGORITHM recursive
+ 	setenv --test GIT_TEST_FULL_IN_PACK_ARRAY true
+ 	setenv --test GIT_TEST_OE_SIZE 10
+ 	setenv --test GIT_TEST_OE_DELTA_SIZE 5
+ 	setenv --test GIT_TEST_COMMIT_GRAPH 1
+ 	setenv --test GIT_TEST_COMMIT_GRAPH_CHANGED_PATHS 1
+ 	setenv --test GIT_TEST_MULTI_PACK_INDEX 1
+ 	setenv --test GIT_TEST_MULTI_PACK_INDEX_WRITE_BITMAP 1
+ 	setenv --test GIT_TEST_ADD_I_USE_BUILTIN 1
+ 	setenv --test GIT_TEST_DEFAULT_INITIAL_BRANCH_NAME master
+ 	setenv --test GIT_TEST_WRITE_REV_INDEX 1
+ 	setenv --test GIT_TEST_CHECKOUT_WORKERS 2
+ 	;;
+ osx-gcc)
+ 	MAKEFLAGS="$MAKEFLAGS PYTHON_PATH=$(which python3)"
 -	CC=gcc
++	CC=gcc-9
+ 	;;
+ osx-clang)
+ 	MAKEFLAGS="$MAKEFLAGS PYTHON_PATH=$(which python2)"
+ 	CC=clang
+ 	;;
+ linux-clang)
+ 	CC=clang
+ 	setenv --test GIT_TEST_DEFAULT_HASH sha1
+ 	;;
+ linux-sha256)
+ 	CC=clang
+ 	setenv --test GIT_TEST_DEFAULT_HASH sha256

ci/lib.sh

Quoted 10 lines
ci/lib.sh +7 -0
+ 	;;
+ pedantic)
+ 	CC=gcc
+ 	# Don't run the tests; we only care about whether Git can be
+ 	# built.
+ 	setenv --build DEVOPTS pedantic
+ 	;;
  linux32)
  	CC=gcc
  	;;

This matches the resolution I have for it. noted here:
https://lore.kernel.org/git/[email protected]/

I.e. it was just those CC=gcc lines for those two jobs that needed to be tweaked to have CC=gcc-8 and CC=gcc-9, respectively. Then likewise for linux-TEST-vars (not in my version, since that was against the v2 of the series, but correct here).

Thanks!