Allow users to choose linker like lld to make build faster.

[jiayuehua]

What type of PR is this?

• bug
• [* ] feature
• enhancement

What does this PR do?

user can chose linker like lld to make build more faster.

Which issue(s)/PR(s) this PR relates to?

none

Special notes for your reviewer, ex. impact of this fix, etc:

If ci can install lld before build in the future, then can make lld default linker. Now still use bfd as default.

Additional context:

none

Checklist：

• [* ] Documentation affected （Please add the label if documentation needs to be modified.)
• Incompatible （If it is incompatible, please describe it and add corresponding label.）
• Need to cherry-pick （If need to cherry-pick to some branches, please label the destination version(s).)
• Performance impacted: Consumes more CPU/Memory

user in the future need to install lld first to link faster.

Release notes：

Please confirm whether to reflect in release notes and how to describe:

                                                            `

support lld linker and make build a litter faster

[Sophie-Xie]

[* ] Documentation affected

--> Does this PR affect the document?

[jiayuehua]

[* ] Documentation affected

--> Does this PR affect the document?

Yes . In the future, user need to install lld first. And we may set lld as default linker to build fast.

[yixinglu]

I think you shouldn't change the linker config, just by passing the NEBULA_USE_LINKER into cmake in configuring stage.

please see the .github/workflows/pull_request.yml content

[jiayuehua]

I think you shouldn't change the linker config, just by passing the NEBULA_USE_LINKER into cmake in configuring stage.

please see the .github/workflows/pull_request.yml content

No, the linker config is explicit telling what kind of linker is supported. If set it to CI, Many developer will not notice it, Also Then Can Set NEBULA_USE_LINKER to wrong value like 'notvalidld'. Set it in linkerconfig file developer can know what kind of linker supported.

[yixinglu]

I think you shouldn't change the linker config, just by passing the NEBULA_USE_LINKER into cmake in configuring stage.
please see the .github/workflows/pull_request.yml content

No, the linker config is explicit telling what kind of linker is supported. If set it to CI, Many developer will not notice it, Also Then Can Set NEBULA_USE_LINKER to wrong value like 'notvalidld'. Set it in linkerconfig file developer can know what kind of linker supported.

make sense. And only thing needing to do is just to verify the compiling in multi-platform.

[codecov-commenter]

Codecov Report

Merging #3460 (11c7704) into master (53b85dc) will increase coverage by 0.02%.
The diff coverage is 75.12%.

@@            Coverage Diff             @@
##           master    #3460      +/-   ##
==========================================
+ Coverage   85.19%   85.22%   +0.02%     
==========================================
  Files        1306     1307       +1     
  Lines      122158   122244      +86     
==========================================
+ Hits       104078   104179     +101     
+ Misses      18080    18065      -15     

Impacted Files	Coverage Δ
src/common/expression/VariableExpression.cpp	84.84% <ø> (-0.45%)	⬇️
...rc/graph/executor/query/AppendVerticesExecutor.cpp	98.30% <ø> (ø)
src/graph/executor/query/GetEdgesExecutor.cpp	94.00% <ø> (ø)
src/graph/executor/query/GetVerticesExecutor.cpp	96.77% <ø> (ø)
src/graph/executor/query/IndexScanExecutor.cpp	88.88% <ø> (ø)
src/graph/optimizer/rule/IndexScanRule.h	100.00% <ø> (ø)
...mizer/rule/PushLimitDownEdgeIndexRangeScanRule.cpp	20.00% <0.00%> (-0.69%)	⬇️
...raph/optimizer/rule/PushLimitDownIndexScanRule.cpp	14.81% <0.00%> (-0.57%)	⬇️
...imizer/rule/PushLimitDownTagIndexRangeScanRule.cpp	20.00% <0.00%> (-0.69%)	⬇️
src/graph/planner/plan/Query.h	97.22% <ø> (ø)
... and 70 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 9ea4889...11c7704. Read the comment docs.