← Back to team overview

yade-dev team mailing list archive

Re: New Yade version, beginning of January 2021 - Release notes.

 

Hi Janek,

thanks for preparing release notes based on MR! Very appreciated!
I will start to prepare a newer Yade release soon.

Best regards

Anton

Am Fr., 25. Dez. 2020 um 00:26 Uhr schrieb Janek Kozicki (yade)
<jkozicki-yade@xxxxxxxxx>:
>
> Hi Anton,
>
> the website https://pad.systemli.org/p/yade-2021-release-notes
> has reset its content. I pasted there the merge requests from this year.
> But also I paste them here. Of course please rearrange as you see fit :)
>
> I might have misplaced some merge request titles. Then please edit this list.
> Especially I thought that we had more MPI related stufff happening.
>
> ============================================
> Short summary of 2021.01a release:
> short title (Merge Request count)
>
> Documentation                         (41 MRs)
> Bug fixes                             (36 MRs)
> High precision RealHP<N>              (31 MRs)
> Small improvements                    (22 MRs)
> Fluid & thermal & particle saturation (14 MRs)
> Tests & checks                        (12 MRs)
> OpenMPI calculations                  ( 8 MRs)
> clang-format in the pipeline          ( 7 MRs)
> Packaging improvements                ( 6 MRs)
> GUI                                   ( 3 MRs)
>
>
> Documentation (41 MRs):
>   Update installation.rst (asan builds + stuff)
>   Add recent publications
>   point youtube's 'related videos' to the same channel
>   Update front page link to videos: tutorial-more-examples.html more up to date.
>   Fix bug and documentation FlowEngine invVoidVolume wrapper
>   Fix recent build failures on master with latest docker bullseye image.
>   Small docs fix
>   Doc typo: missing norm in trial shear force
>   Doc Body.groupMask
>   User doc
>   Update high precision documentation
>   Miscellanea doc
>   Add example script for opengeomechanics paper
>   clarify thermalengine doc
>   Documentation makeCloud and makeClumpCloud
>   ViscoelasticPM doc update
>   Prog doc typo
>   ProgDocTypos
>   Formulation doc typo
>   improve cellDist doc
>   small addition to the doc to mention verletDist explicitely
>   Doc prog.rst update
>   FlowEngine.getCell doc
>   Doc update of VTKRecorder.fileName
>   ymport.textPolyhedra-doc
>   Prog doc typos and minor changes
>   Doc Law2_Polyhedra*
>   residualFrictionAngle in radians in JCFpmMat docstring
>   Sphinx doc broken link in Prog doc
>   O.forces.f doc
>   Two Docstrings
>   BaselineTypo
>   adding open geomechanics reference to doc and pubs
>   Add --stdperformance, --quickperformance explanation to documentation.
>   Polyhedra Ip2 doc
>   MaterialContainer doc typo
>   timing.runtimeDoc
>   Added Hicham Benniou PhD reference
>   Add references to article describing lubrication law
>   update publications
>   !463 alternative : Added article February 2020 Computational Particle Mechanics
>
> Bug fixes (36 MRs):
>   fix #145 seed in makeCloud FIX makeCloud doesn't use "seed"
>   Fix pygts module import in python3.9
>   Fix PyRunner. Now it can propagate errors upwards.
>   Fix #110 FIX Graph yade.plot.live matplotlib numpy update exception
>   Fix compilation with python3.9
>   fix #157 FIX O.stopAtTime does not do anything useful
>   Fix #140 FIX ForceContainerParallel::sync() can count permanent forces twice or more
>   Fix https://gitlab.com/yade-dev/docker-yade/issues/1 FIX yade-ci crashes
>   export.text2vtk - fixed a bug + refactoring
>   Fix #146 FIX "to-Python converter already registered" RuntimeWarning
>   Fix even more warnings, print only 'maybe-uninitialized' and '#warning' (!270), faster compilation. (!273)   18 of 18 tasks completed
>   fix clang formating mistake, change remaining doubles to Real
>   fix tutorial script https://answers.launchpad.net/yade/+question/689537
>   Fix printing Vectors by LOG macros.
>   Fix GCC-10 compilation #958006
>   Fix_PBs_orientations
>   #153 fix FIX O.energy['nonViscDamp'] misses gravity forces
>   Skip url ping-pong for "getting help" page
>   fix yade crash, when closing 3D view via "X" window frame button, in Ubuntu 20.04, fixes #164 FIX yade crash when closing the 3D view.
>   Fix close view crash
>   small mailmap fix and test cleanup
>   Little bit of work on #97 : lib/base/Math.hpp should be math only
>   Fix compilation against newer QT5.
>   Some small fixes for Sacha : Fix compilation with boost 1.65 and g++ 9.3 combination
>   Fix denser grid in GLViewer.
>   clumpId type should be Body::id_t, not int. Add fixme note in Math.hpp about issue #97
>   Fix #162 - fix building master pipeline due to recent VTK upgrade in OpenSuse 15.
>   Fix build on latest bullseye
>   Fix doc generation for newer sphinx. #955097
>   Workaround MPFR boost bug on ubuntu focal
>   Enable OpenMP for MPFR
>   SpatialQuickSortCollider is nice, because it's only 70 lines long. But has one bug. FIX groupMask treatment.
>   comment out buggy heat flux boundary condition for now
>   GCAL version 5.1.1 produces different results (again), but maybe the old results are back?
>   Qt::MidButton is deprecated, use MiddleButton instead
>   Use boost::filter_iterator for BodyContainer instead of home-backed. #122
>
> High precision RealHP<N> (31 MRs):
>   Finalize high precision Real - finished, really :) documentation + put minieigen sources into Yade
>   Add check test for Runge Kutta Cash Karp 54 time integrator, error tolerance 1e-147, 147 correct decimal places :)
>   checkGravity.py : reduce error tolerance from 1e-3 → 1e-10
>   Use the minieigen ↔ yade.minieigenHP redirection when necessary.
>   Add High Precision to the gitlab CI pipeline
>   Ensure that VTK is compatibile with Real - Without Macros.
>   std::weibull_distribution can't understand Real type so use boost::weibull_distribution
>   OpenGL Real compatibility
>   Lapack Compatibility with Real type.
>   Remove Real from global namespace.
>   Another step towards enabling high precision Real
>   Adapt checks and tests to high precision Real.
>   Add RealHP<N> : double, quadruple and more precisions.
>   Bit correctness: extra tests for RealHP<N>
>   Small HP cleanup: don't use std::to_string.
>   The last remaining double to Real changes (2/2)
>   The last remaining double to Real changes (1/2)
>   Remove #DEFINEs from MathFunctions.hpp
>   Remove unused variables uN, uT.
>   don't use operator++ for floating point type.
>   Replace std:: with math:: when invoking math functions. (6/4) - optional, remove even more using std::min, std::max.
>   Replace std:: with math:: when invoking math functions. (2/4)
>   Replace std:: with math:: when invoking math functions. (3/4)
>   Replace std:: with math:: when invoking math functions. (4/4)
>   Replace std:: with math:: when invoking math functions. (5+6/4) - optional, remove more using
>   Replace std:: with math:: when invoking math functions. (1/4)
>   put minieigen sources back into Yade
>   Remove a couple of non-math:: using directives.
>   If necessary use mpmath.linspace instead of numpy.linspace. (high precision on python side)
>   Enable high precision CGAL
>   Avoid raw memory access for high precision types.
>
> Small improvements (22 MRs):
>   added utils.randomOrientation function
>   libVersions detects architecture.
>   Use std::ofstream in pkg/dem/PDFEngine.cpp instead of old C interface for writing data files.
>   Use C++17 standard, remove Loki dependency.
>   Use vtkSmartPointer to handle memory release
>   small mailmap update.
>   Merge PotentialBlockVTKRecorder and PotentialBlockVTKRecorderTunnel engines
>   A better Wall-Polyhedra normal
>   Update warning message for deprecated GravityEngine
>   Clean shape: Add #ifdef YADE_FEM in core/Shape.hpp
>   Add seed value for periodic packings
>   improved O.stopAtTime
>   frictionAngle in radians for examples
>   exportPotentialBlocks
>   Removing Vector4r definition from Lin4NodeTetra
>   Add ymport.stl.shift
>   Add ymport.stl.scale parameter
>   change Shop::aabbExtrema from a Python tuple to a C++ pair.
>   Add seed parameter to MakeClumpCloud
>   Un-expose breakOccurred to avoid user confusion
>   remove remaining FOREACH loops on bodies
>   consistent use of 'et' in ViscoelasticPM (LP Answer 691009)
>
> Fluid & thermal & particle saturation (14 MRs):
>   Adding flux type boundary condition (thermalEngine) for Luc Scholtes and Ahmed Mostafa
>   Protect from segfault after looping on erased particles in thermalengine
>   improving the particle sat field estimate, other small improvements
>   Partial sat clay patch
>   Partial sat update
>   add Ip2 functor for PartialSatMat
>   Add lubrication examples
>   add the packing used in thermalengine examples
>   add example scripts demonstrating ThermalEngine functionality
>   decouple mechanical and partialsat timesteps
>   Update conductionVerification.py
>   add fully coupled THM example, add citation to thermalengine publication
>   a more robust cmake when cgal or pfvflow are disabled and dependent modules...
>   add PFV tools for getting cell velocity and volume
>
> Tests & checks (12 MRs):
>   Improve multithread debugging and logging. Add TRVARn(…), LOG_TIMED_*(5seconds,msg)
>   Make sure gitlab CI pipeline shows all the output.
>   improve error checking
>   Print time spent on each of the --checks
>   Add VTKRecorder check.
>   Add vectorized compilation to the pipeline
>   checkPotentialVTKRecorders
>   make runtime of two checkScripts acceptable
>   Improve performance test, add it to the pipeline.
>   Add multi core check
>   add enum tests and fix .deb builds
>   Add 32bit architecture to the pipeline.
>
> OpenMPI calculations (8 MRs):
>   Correct and improve two methods in domaindecomposition.py
>   Mpy update
>   cleaner exit and spawn/despawn
>   MpiDoc rebased
>   One Py3-required LOG_TRACE, Making Py3 mandatory for MPI compilation. Fix #144 Py2 (no-)support
>   fix thread-unsafe assignment of permForceSynced (fix #156, setPermF broken)
>   Temporarily disable insertionSortParallel() until #138 is solved.
>   fix parallel collision detection (fix #138) : FIX Out-of-bounds access in insertionSortParallel
>
> clang-format in the pipeline (7 MRs):
>   clang-format ./pkg
>   clang-format ./py
>   clang-format ./lib
>   clang-format ./core
>   clang-format gui/qt5. Remove qt4.
>   Add clang-format test to the CI pipeline.
>   clang-format discussion.
>
> Packaging improvements (6 MRs):
>   Prepare 20.04 deb-packages
>   Add instructions for yadedaily Ubuntu 20.04 packages
>   Update scripts for uploading daily-packages
>   Fix deb packages
>   build deb also on master pipeline
>   Clean aptly DB after update
>
> GUI (3 MRs):
>   Draw grid label coordinates, allow custom change of grid density.
>   remove unaligned margins in GUI inspect.
>   Add enum support for GUI dropdown menus.
>
> ===
> Are the extra debian packages yade-long-double or yade-float128, maybe even yade-mpfr150
> even possible?
>
> best regards
> Janek
>
> Janek Kozicki (yade) said:     (by the date of Sun, 6 Dec 2020 00:26:03 +0100)
>
> > Hi Anton,
> >
> > Do you think that you could also prepare yade-long-double or
> > yade-float128, maybe even yade-mpfr150 packages?
> >
> > Better if all of them used ENABLE_MPFR, to avoid slow boost::cpp_bin_float.
> >
> > It would work great to protect (via dpkg tests caused by changes in
> > other packages) our achievement of having working high precision in
> > yade. We don't want this functionality to break at some later time.
> >
> > best regards
> > Janek
> >
> >
> > Anton Gladky said:     (by the date of Mon, 16 Nov 2020 20:13:06 +0100)
> >
> > > Hello Jérôme,
> > >
> > > our goal is to get the newer Yade into the Debian Bullseye, because
> > > it will be the default version for the next two years (till 2023).
> > >
> > > It would be good to upload Yade into the repository in January (till soft freeze
> > > on 2021-02-12). After that date any upload into the repository should be
> > > coordinated with the Release team and is actually not for the new software
> > > versions.
> > >
> > > If you are able to push your changes till the end of December and all tests
> > > are passing, I think it would be possible to accept it for the new release.
> > >
> > > I am not a fan of a last-minute commit (especially on Friday afternoon
> > > :) ) because it can break the stuff with a very limited time to fix it.
> > >
> > > Anyway, if you are not able to do it on time - there is still an
> > > opportunity to release a minor version later.
> > >
> > > Please take time and test your code (including --test and --check scripts)
> > > to prevent breakages and to provide Yade with new and reliable features.
> > >
> > > Best regards
> > >
> > >
> > > Anton
> > >
> > > Am Fr., 13. Nov. 2020 um 11:23 Uhr schrieb Jerome Duriez
> > > <jerome.duriez@xxxxxxxx>:
> > >
> > > >
> > > > Dear all,
> > > >
> > > > Sorry for the inconvenience but is there any flexibility in that schedule ?
> > > >
> > > > I have been working on a new Shape child and I may get closer to the
> > > > endpoint of my work in the form of a merge request.
> > > >
> > > > It now represents "32 files changed, 2193 insertions(+), 178
> > > > deletions(-)" since 2020.01(a ?) = commit 9964f5.
> > > >
> > > >
> > > > In case there is some flexibility in the schedule and if I'm not the
> > > > only one interested in that flexibility (MPI maybe ?..), maybe we could
> > > > discuss inclusion of that Shape child in the new release ?
> > > >
> > > > If not, let it be, I would totally understand to miss that train.
> > > >
> > > >
> > > > Jérôme
> > > >
> > > > ------
> > > > Chargé de Recherche / Research Associate
> > > > INRAE, RECOVER
> > > > 3275 route Cezanne – CS 40061 13182 Aix-en-Provence Cedex 5 FRANCE
> > > > +33 (0)4 42 66 99 21
> > > > https://www6.paca.inrae.fr/recover/membres-du-laboratoire/pages-personnelles/jerome-duriez
> > > >
> > > > On 12/11/2020 22:18, Anton Gladky wrote:
> > > > > Dear Yade users and developers,
> > > > >
> > > > > As always at the beginning of January I am planning to prepare
> > > > > a release of a new Yade version.
> > > > >
> > > > > Please try to push your changes inyo the code at least till the mid
> > > > > of December, so we will have enough time to test it on different
> > > > > platforms during the Christmas period and release it on time.
> > > > > Also please try not to push too much breaking changes at that
> > > > > period..
> > > > >
> > > > > The new Debian Bullseye release is scheduled already [1]. So we
> > > > > should not be too late to prepare a new Yade for the next stable
> > > > > Debian version.
> > > > >
> > > > > Also it would be good to prepare short but meaningful release notes.
> > > > > Current git-workflow has too many small log-messages, so it is
> > > > > difficult to get meaningful information from there. Please use this
> > > > > link to add some notes [2].
> > > > >
> > > > > [1] https://wiki.debian.org/DebianBullseye
> > > > > [2] https://pad.systemli.org/p/yade-2021-release-notes
> > > > >
> > > > > Thank you
> > > > >
> > > > > Anton Gladky
> --
> Janek Kozicki, PhD. DSc. Arch. Assoc. Prof.
> Gdańsk University of Technology
> Faculty of Applied Physics and Mathematics
> Department of Theoretical Physics and Quantum Information
> --
> http://yade-dem.org/
> http://pg.edu.pl/jkozicki (click English flag on top right)
>
> _______________________________________________
> Mailing list: https://launchpad.net/~yade-dev
> Post to     : yade-dev@xxxxxxxxxxxxxxxxxxx
> Unsubscribe : https://launchpad.net/~yade-dev
> More help   : https://help.launchpad.net/ListHelp


Follow ups

References