| FazBrowse GitHub Viewer | Trending | | Home |
| Tools: [Download Repo ZIP] [Original HTTPS Page] |
Sorry, something went wrong.
|
@refack build started: https://ci.nodejs.org/blue/organizations/jenkins/node-test-pull-request-lite-pipeline/detail/node-test-pull-request-lite-pipeline/1309/pipeline |
Sorry, something went wrong.
|
/CC @nodejs/build-files @nodejs/testing |
Sorry, something went wrong.
There was a problem hiding this comment.
Refs: #23733
Sorry, something went wrong.
There was a problem hiding this comment.
I’d prefer to undo this unless it’s actually causing trouble on Travis – I don’t think we’ve seen this so far.
Sorry, something went wrong.
There was a problem hiding this comment.
It makes the tests consistent with JEnkins, but I'll do it in another PR
Sorry, something went wrong.
There was a problem hiding this comment.
I’d prefer to undo this unless it’s actually causing trouble on Travis – I don’t think we’ve seen this so far.
Sorry, something went wrong.
There was a problem hiding this comment.
Maybe set these using export CC='ccache gcc-4.9' && CXX='ccache g++-4.9' on the first line of install:, and then leave these lines as they are?
Also, I’d prefer to keep ./configure – firstly, the .travis.yml of a project is sometimes used by devs to figure out how to build it, because it’s usually much more to-the-point than build instructions, and secondly that way we simply test more code than without it (namely the contents of configure).
Sorry, something went wrong.
There was a problem hiding this comment.
I'm suspicious it's not behaving as we expect (hence it's using GCC over clang), so I'd rather has it explicit (like in Jenkins).
Ack on the ./configure
Sorry, something went wrong.
There was a problem hiding this comment.
Testing export in install
Sorry, something went wrong.
From a quick google search, it seems like the error message is a clang one. |
Sorry, something went wrong.
So it might be just an stdlibc++ issue? |
Sorry, something went wrong.
|
@refack My guess is that it might be the gcc-provided libstdc++ 4.8 not playing along well with clang? Could be totally wrong on that one, though. (I also don’t remember the reason for picking clang in the first place, tbh.) |
Sorry, something went wrong.
|
I also changed the output generator from make -j2 V= make -C out BUILDTYPE=Release V= make[1]: Entering directory `/home/travis/build/nodejs/node/out' CXX(host) /home/travis/build/nodejs/node/out/Release/obj.host/v8_libbase/deps/v8/src/base/bits.o CXX(host) /home/travis/build/nodejs/node/out/Release/obj.host/v8_libbase/deps/v8/src/base/cpu.o CXX(host) /home/travis/build/nodejs/node/out/Release/obj.host/v8_libbase/deps/v8/src/base/debug/stack_trace.o CXX(host) /home/travis/build/nodejs/node/out/Release/obj.host/v8_libbase/deps/v8/src/base/division-by-constant.o CXX(host) /home/travis/build/nodejs/node/out/Release/obj.host/v8_libbase/deps/v8/src/base/file-utils.o to $ make -j2 V=1 | sed -e "s/'.*\$//" -e "s|$PWD||g" make -C out BUILDTYPE=Release V=1 make[1]: Entering directory `/out ccache g++-4.9 -o /out/Release/obj.host/v8_libbase/deps/v8/src/base/bits.o ../deps/v8/src/base/bits.cc ccache g++-4.9 -o /out/Release/obj.host/v8_libbase/deps/v8/src/base/cpu.o ../deps/v8/src/base/cpu.cc ccache g++-4.9 -o /out/Release/obj.host/v8_libbase/deps/v8/src/base/debug/stack_trace.o ../deps/v8/src/base/debug/stack_trace.cc (this is planned to change in Jenkins as well nodejs/build#1517) |
Sorry, something went wrong.
|
@refack One thing I’m noticing here is that the extra apt install took about a minute in https://travis-ci.com/nodejs/node/jobs/153110829 … not the end of the world, but something we might want to consider? |
Sorry, something went wrong.
There was a problem hiding this comment.
It seems nodejs/build#1517 is still being discussed. I'm not a fan of this change, can we revert it?
Sorry, something went wrong.
There was a problem hiding this comment.
ack
Sorry, something went wrong.
But for some reason total was 13m While currently the jobs (using clang) take 15m+ |
Sorry, something went wrong.
|
It's back: In file included from ../src/bootstrapper.cc:2:
In file included from ../src/env-inl.h:27:
../src/aliased_buffer.h:27:27: error: no template named 'enable_if_t' in namespace 'std'; did you mean 'enable_if'?
typename = std::enable_if_t<std::is_scalar<NativeT>::value>>
~~~~~^~~~~~~~~~~
enable_if
/usr/lib/gcc/x86_64-linux-gnu/4.8/../../../../include/c++/4.8/type_traits:1766:12: note: 'enable_if' declared here
struct enable_if
^
1 error generated.
|
Sorry, something went wrong.
Sorry, something went wrong.
The version of `clang` provided in the Travis linux image uses libstdc++4.8 whice is below our minimal supported version. Switching to `make test -j1` is to avoid races during the test cycle causes by the main target being "unstable", that is it always builds some files, and relinks the binary, which is used by the test procedure. PR-URL: nodejs#23778 Reviewed-By: Luigi Pinca <luigipinca@gmail.com> Reviewed-By: Richard Lau <riclau@uk.ibm.com>
The version of `clang` provided in the Travis linux image uses libstdc++4.8 whice is below our minimal supported version. Switching to `make test -j1` is to avoid races during the test cycle causes by the main target being "unstable", that is it always builds some files, and relinks the binary, which is used by the test procedure. PR-URL: #23778 Reviewed-By: Luigi Pinca <luigipinca@gmail.com> Reviewed-By: Richard Lau <riclau@uk.ibm.com>
The version of `clang` provided in the Travis linux image uses libstdc++4.8 whice is below our minimal supported version. Switching to `make test -j1` is to avoid races during the test cycle causes by the main target being "unstable", that is it always builds some files, and relinks the binary, which is used by the test procedure. PR-URL: #23778 Reviewed-By: Luigi Pinca <luigipinca@gmail.com> Reviewed-By: Richard Lau <riclau@uk.ibm.com>
The version of `clang` provided in the Travis linux image uses libstdc++4.8 whice is below our minimal supported version. Switching to `make test -j1` is to avoid races during the test cycle causes by the main target being "unstable", that is it always builds some files, and relinks the binary, which is used by the test procedure. PR-URL: #23778 Reviewed-By: Luigi Pinca <luigipinca@gmail.com> Reviewed-By: Richard Lau <riclau@uk.ibm.com>
The version of `clang` provided in the Travis linux image uses libstdc++4.8 whice is below our minimal supported version. Switching to `make test -j1` is to avoid races during the test cycle causes by the main target being "unstable", that is it always builds some files, and relinks the binary, which is used by the test procedure. PR-URL: #23778 Reviewed-By: Luigi Pinca <luigipinca@gmail.com> Reviewed-By: Richard Lau <riclau@uk.ibm.com>
The version of `clang` provided in the Travis linux image uses libstdc++4.8 whice is below our minimal supported version. Switching to `make test -j1` is to avoid races during the test cycle causes by the main target being "unstable", that is it always builds some files, and relinks the binary, which is used by the test procedure. PR-URL: #23778 Reviewed-By: Luigi Pinca <luigipinca@gmail.com> Reviewed-By: Richard Lau <riclau@uk.ibm.com>
The version of `clang` provided in the Travis linux image uses libstdc++4.8 whice is below our minimal supported version. Switching to `make test -j1` is to avoid races during the test cycle causes by the main target being "unstable", that is it always builds some files, and relinks the binary, which is used by the test procedure. PR-URL: #23778 Reviewed-By: Luigi Pinca <luigipinca@gmail.com> Reviewed-By: Richard Lau <riclau@uk.ibm.com>
The version of `clang` provided in the Travis linux image uses libstdc++4.8 whice is below our minimal supported version. Switching to `make test -j1` is to avoid races during the test cycle causes by the main target being "unstable", that is it always builds some files, and relinks the binary, which is used by the test procedure. PR-URL: #23778 Reviewed-By: Luigi Pinca <luigipinca@gmail.com> Reviewed-By: Richard Lau <riclau@uk.ibm.com>
| Back | FazBrowse Home | New Git URL |
The version of clang provided in the Travis linux image uses
libstdc++4.8 whice is below our minimal supported version.
Switching to make test -j1 is to avoid races during the test cycle
Detailscauses by the main target being "unstable", that is it always builds
some files, and relinks the binary, which is used by the test procedure.
../src/node_crypto.cc:3641:29: error: no matching function for call to 'get' return sign->CheckThrow(std::get<Error>(ret)); ^~~~~~~~~~~~~~~ /usr/lib/gcc/x86_64-linux-gnu/4.8/../../../../include/c++/4.8/utility:142:5: note: candidate template ignored: invalid explicitly-specified argument for template parameter '_Int' get(std::pair<_Tp1, _Tp2>& __in) noexcept ^ /usr/lib/gcc/x86_64-linux-gnu/4.8/../../../../include/c++/4.8/utility:147:5: note: candidate template ignored: invalid explicitly-specified argument for template parameter '_Int' get(std::pair<_Tp1, _Tp2>&& __in) noexcept ^ /usr/lib/gcc/x86_64-linux-gnu/4.8/../../../../include/c++/4.8/utility:152:5: note: candidate template ignored: invalid explicitly-specified argument for template parameter '_Int' get(const std::pair<_Tp1, _Tp2>& __in) noexcept ^ /usr/lib/gcc/x86_64-linux-gnu/4.8/../../../../include/c++/4.8/array:268:5: note: candidate template ignored: invalid explicitly-specified argument for template parameter '_Int' get(array<_Tp, _Nm>& __arr) noexcept ^ /usr/lib/gcc/x86_64-linux-gnu/4.8/../../../../include/c++/4.8/array:277:5: note: candidate template ignored: invalid explicitly-specified argument for template parameter '_Int'