Fix GCC 11 Compiler Warnings #1213

franzpoeschel · 2022-03-02T15:10:25Z

Compile warnings in Attribute.hpp:
g++ 11 does not understand that these variables are not primitive types
and complains about possibly uninitialized values. Use the {}
constructor to fix this.
We still had two instances of AccessType in the tests, replace this with Access

ax3l · 2022-03-07T17:31:52Z

test/SerialIOTest.cpp

        std::unique_ptr<openPMD::Series> series_ptr;
        {
-            openPMD::Series series(
-                "sample%T.json", openPMD::AccessType::CREATE);


We left those intentionally to test that the old type still works.

We could silence this at one last location with a local suppressor?

I was under the impression that they sneaked in because this test comes from a long-standing pull request (the bug tested by this took a long time to fix).
But we can definitely use the chance to test for existence of AccessType with this

ax3l · 2022-03-07T18:32:48Z

PR rebased.

G++ 11 does not understand that these variables are not primitive types and complains about possibly uninitialized values. Use the {} constructor to fix this.

franzpoeschel · 2022-03-15T17:25:36Z

include/openPMD/backend/Attribute.hpp

    }

-    (void)pv;
    throw std::runtime_error("getCast: no cast possible.");


Need to put these into an else branch, ICC seems to determine unreachable statements after resolving if constexpr

ah classic. a range of NVCC 11 minor releases does that as well

ICC seems to resolve reachability of statements after resolving constexpr if

test/SerialIOTest.cpp

include/openPMD/backend/Attribute.hpp

1. Reset to default warning behavior after disabling 2. Use __GNUC_MINOR__ instead of __GNUC__

ax3l · 2022-03-29T06:38:39Z

test/SerialIOTest.cpp

+     */
+#if defined(__GNUC__MINOR__) || defined(__clang__)
+#pragma GCC diagnostic push
+#pragma GCC diagnostic ignored "-Wdeprecated-declarations"


With this one set, you can remove -Wno-deprecated-declarations from quite a few CI scripts in .github/workflows/

ax3l · 2022-04-26T03:31:20Z

test/SerialIOTest.cpp

+    /*
+     * clang also understands these pragmas.
+     */
+#if defined(__GNUC__MINOR__) || defined(__clang__)


Typo fixed in #1255: __GNUC_MINOR__

ax3l · 2022-04-26T03:31:41Z

test/SerialIOTest.cpp

+#if defined(__INTEL_COMPILER)
+#pragma warning(disable : 2282)
+#endif
+#if defined(__GNUC__MINOR__) || defined(__clang__)


There was a typo in `__GNUC_MINOR__`, causing the test not to work on all GCC versions. Follow-up to openPMD#1213

There was a typo in `__GNUC_MINOR__`, causing the test not to work on all GCC versions. Follow-up to #1213

ax3l · 2022-05-25T16:32:00Z

I have problems applying this and the following PRs to the release-0.14.5 branch

There was a typo in `__GNUC_MINOR__`, causing the test not to work on all GCC versions. Follow-up to openPMD#1213

* Don't safeguard empty strings while reading * Update TODO Comment * Add missing update for VEC_CLONG_DOUBLE * Ignore deprecated AccessType in SerialIOTest * SerialIOTest: Fix GCC Pragma Check (openPMD#1260) There was a typo in `__GNUC_MINOR__`, causing the test not to work on all GCC versions. Follow-up to openPMD#1213 * Clearly fail when users select a wrong backend * ICC: Unreachable end of function * Fix use after free in ADIOS1IOHandler * Remove unavailable_backend_test Hard to trigger without explicit backend selection, and test coverage is present on dev anyway Co-authored-by: Axel Huebl <[email protected]>

ax3l · 2022-06-06T22:04:43Z

Applied to 0.14.5 via ax3l#6 🎉

* Improve write time * Print warning if mpi4py is not found in openpmd-pipe * ADIOS1: Remove task from IO queue if it fails with exception * `setup.py`: Extra CMake Arg Control * EMSCRIPTEN: Skip Py Extension * g++/clang++: Add `-Wsign-compare` * python/RecordComponent: fix `-Wsign-compare` * Patch MSVC pybind11 bug Improve patch comment * Python: Do Not Strip Symbols In Debug Avoid stripping symbols for Python debug builds, so we can see lines in coredumps and debugger runs. * CI: NVHPC New Apt Repo Update the NVHPC install instructions to the latest and greatest. Fix failing CI (dependency install). Also upgrade to 21.11 as on Perlmutter. * clang-format & pre-commit & python include * run pre-commit * ps_make_timer_name_: add memory leak suppression * Fix: Missing HDF5 Include This tries to fix a compile error on Maxwell (DESY). ``` /home/diederse/software/hipace_4GPU/hipace/build/_deps/fetchedopenpmd-src/src/IO/HDF5/HDF5IOHandler.cpp: In member function 'virtual void openPMD::HDF5IOHandlerImpl::readAttribute(openPMD::Writable*, openPMD::Parameter<openPMD::Operation::READ_ATT>&)': /home/diederse/software/hipace_4GPU/hipace/build/_deps/fetchedopenpmd-src/src/IO/HDF5/HDF5IOHandler.cpp:1586:17: error: 'H5free_memory' was not declared in this scope 1586 | H5free_memory(m1); | ^~~~~~~~~~~~~ /home/diederse/software/hipace_4GPU/hipace/build/_deps/fetchedopenpmd-src/src/IO/HDF5/HDF5IOHandler.cpp:1609:17: error: 'H5free_memory' was not declared in this scope 1609 | H5free_memory(m1); | ^~~~~~~~~~~~~ /home/diederse/software/hipace_4GPU/hipace/build/_deps/fetchedopenpmd-src/src/IO/HDF5/HDF5IOHandler.cpp:1628:17: error: 'H5free_memory' was not declared in this scope 1628 | H5free_memory(m2); | ^~~~~~~~~~~~~ ``` * Increase reference count also in other load_chunk overload Don't know if it is necessary, but looks like we forgot it earlier Add comment * Spack: Remove Old Files * Spack: v0.17.1 * Update Env Files * Python Iteration: Fix __repr__ (time) Small numbers, as common for iterations, were flushed to zero in `std::to_string(double)` of the representation of `Iteration` variables: ``` step __repr__ 100 <openPMD.Iteration at t = '0.000000 s'> was: 8.687655225973454e-14 * 1.0 ``` * CI: Update CUDA repo key Nvidia has made changes in the signing keys. https://forums.developer.nvidia.com/t/notice-cuda-linux-repository-key-rotation/212771 * CI: Switch to Mamba * Doc: Update HDF5 Coll. Metadata Versions Update the documentation about HDFFV-11260 and the newly released fixes in release lines. * HDF5 Coll. MD Reads: Simplify Wording * Tiny wording improvement and a + * Remove deprecated debug parameter in ADIOS2 * HDF5IOHandler: Clang-Format/Clang-Tidy ``` /home/runner/work/openPMD-api/openPMD-api/src/IO/HDF5/HDF5IOHandler.cpp:1835:21: warning: repeated branch in conditional chain [bugprone-branch-clone] isLegacyLibSplashAttr = ^ ``` * `conda.yaml`: add `pre-commit` * Fix: Python Variant Issue on Conda * Test: Demonstrate Pattern Issue As in #1173 * JSON: Improve File Open Error Message Include path to file * Upon parsing, store each iteration's filename If the padding is inconsistent, a later Iteration::open() needs the original filename. Trying to compute the filename from the expansion pattern will lead to wrong filenames. * Backport: fix-open-iteration (#5) * Pass-through flushing parameters * CI fixes * Test * Don't flush when opening an iteration * CI fixes * FlushLevel: Use default base class in NVC++ * clang-tidy: Define member defaults of Writable in-class * Update include/openPMD/IO/AbstractIOHandler.hpp Co-authored-by: Axel Huebl <[email protected]> * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * Release 0.14.5 conflicting backports (#6) * Don't safeguard empty strings while reading * Update TODO Comment * Add missing update for VEC_CLONG_DOUBLE * Ignore deprecated AccessType in SerialIOTest * SerialIOTest: Fix GCC Pragma Check (#1260) There was a typo in `__GNUC_MINOR__`, causing the test not to work on all GCC versions. Follow-up to #1213 * Clearly fail when users select a wrong backend * ICC: Unreachable end of function * Fix use after free in ADIOS1IOHandler * Remove unavailable_backend_test Hard to trigger without explicit backend selection, and test coverage is present on dev anyway Co-authored-by: Axel Huebl <[email protected]> * Version: 0.14.5 including Changelog Co-authored-by: Jean Luca Bez <[email protected]> Co-authored-by: Franz Pöschel <[email protected]> Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

franzpoeschel added warning cosmetic internal labels Mar 2, 2022

ax3l added this to the 0.14.5 milestone Mar 7, 2022

ax3l reviewed Mar 7, 2022

View reviewed changes

ax3l self-assigned this Mar 7, 2022

ax3l changed the title ~~Fix some compile warnings~~ Fix GCC 11 Compiler Warnings Mar 7, 2022

ax3l force-pushed the fix-compile-warnings branch from bae466f to 2d68c79 Compare March 7, 2022 18:32

franzpoeschel force-pushed the fix-compile-warnings branch from 2d68c79 to 62a4c4c Compare March 8, 2022 12:05

franzpoeschel added 3 commits March 8, 2022 13:18

Compile warnings in Attribute.hpp

1345a07

G++ 11 does not understand that these variables are not primitive types and complains about possibly uninitialized values. Use the {} constructor to fix this.

Ignore deprecated AccessType in SerialIOTest

f2ea3e1

ICC issue: statement unreachable

8d736b4

franzpoeschel force-pushed the fix-compile-warnings branch from 62a4c4c to 8d736b4 Compare March 8, 2022 12:18

franzpoeschel commented Mar 15, 2022

View reviewed changes

franzpoeschel added 3 commits March 15, 2022 18:26

Put throw statement into an else branch

2d29dd3

ICC seems to resolve reachability of statements after resolving constexpr if

Add pragma for ICPC

697d657

Push pragma only in G++ and Clang

6aba021

franzpoeschel force-pushed the fix-compile-warnings branch from 4904c90 to 6aba021 Compare March 21, 2022 10:32

ax3l reviewed Mar 22, 2022

View reviewed changes

test/SerialIOTest.cpp Outdated Show resolved Hide resolved

ax3l reviewed Mar 22, 2022

View reviewed changes

include/openPMD/backend/Attribute.hpp Show resolved Hide resolved

Two fixes

0804165

1. Reset to default warning behavior after disabling 2. Use __GNUC_MINOR__ instead of __GNUC__

franzpoeschel requested a review from ax3l March 24, 2022 08:48

ax3l reviewed Mar 29, 2022

View reviewed changes

ax3l approved these changes Apr 15, 2022

View reviewed changes

ax3l merged commit 88c3ad1 into openPMD:dev Apr 15, 2022

ax3l mentioned this pull request Apr 15, 2022

CI: No Need for Warn Deprecations #1246

Merged

ax3l mentioned this pull request Apr 26, 2022

SerialIOTest: Fix GCC Pragma Check #1255

Merged

ax3l reviewed Apr 26, 2022

View reviewed changes

ax3l added a commit to ax3l/openPMD-api that referenced this pull request Apr 29, 2022

SerialIOTest: Fix GCC Pragma Check

2ba0ca7

There was a typo in `__GNUC_MINOR__`, causing the test not to work on all GCC versions. Follow-up to openPMD#1213

ax3l mentioned this pull request Apr 29, 2022

SerialIOTest: Fix GCC Pragma Check #1260

Merged

ax3l added a commit to ax3l/openPMD-api that referenced this pull request Apr 29, 2022

SerialIOTest: Fix GCC Pragma Check

7fb039b

There was a typo in `__GNUC_MINOR__`, causing the test not to work on all GCC versions. Follow-up to openPMD#1213

ax3l added a commit that referenced this pull request Apr 29, 2022

SerialIOTest: Fix GCC Pragma Check (#1260)

861f719

There was a typo in `__GNUC_MINOR__`, causing the test not to work on all GCC versions. Follow-up to #1213

ax3l modified the milestones: 0.14.5, 0.14.6 May 25, 2022

franzpoeschel pushed a commit to franzpoeschel/openPMD-api that referenced this pull request May 30, 2022

SerialIOTest: Fix GCC Pragma Check (openPMD#1260)

88d48b1

There was a typo in `__GNUC_MINOR__`, causing the test not to work on all GCC versions. Follow-up to openPMD#1213

franzpoeschel mentioned this pull request May 30, 2022

Release 0.14.5 conflicting backports ax3l/openPMD-api#6

Merged

franzpoeschel pushed a commit to franzpoeschel/openPMD-api that referenced this pull request May 31, 2022

SerialIOTest: Fix GCC Pragma Check (openPMD#1260)

73df696

There was a typo in `__GNUC_MINOR__`, causing the test not to work on all GCC versions. Follow-up to openPMD#1213

ax3l modified the milestones: 0.14.6, 0.14.5 Jun 6, 2022

Fix GCC 11 Compiler Warnings #1213

Fix GCC 11 Compiler Warnings #1213

Uh oh!

Conversation

franzpoeschel commented Mar 2, 2022

Uh oh!

ax3l Mar 7, 2022

Choose a reason for hiding this comment

Uh oh!

franzpoeschel Mar 8, 2022

Choose a reason for hiding this comment

Uh oh!

ax3l commented Mar 7, 2022

Uh oh!

franzpoeschel Mar 15, 2022

Choose a reason for hiding this comment

Uh oh!

ax3l Mar 29, 2022

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

ax3l Mar 29, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ax3l Apr 15, 2022

Choose a reason for hiding this comment

Uh oh!

ax3l Apr 26, 2022

Choose a reason for hiding this comment

Uh oh!

ax3l Apr 26, 2022

Choose a reason for hiding this comment

Uh oh!

ax3l commented May 25, 2022

Uh oh!

ax3l commented Jun 6, 2022

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

ax3l Mar 29, 2022 •

edited

Loading