lp:~asc/fluidity/fixup-version-number2

Branch merges

Propose for merging

No branches dependent on this one.

Merged into lp:fluidity at revision 3571

Tim Bond (community): Approve on 2011-08-22

Diff: 18 lines (+7/-1)

1 file modified

Makefile.in (+7/-1)

api

Related source package recipes

No recipes using this branch. (?)

Create packaging recipe

Related snap packages

Related bugs

Bug #814839: Revision number in usage not updated without a 'make clean'

Low

Fix Released

Link a bug report

Related blueprints

3565. By Adam Candy on 2011-08-17

This forces the version number to be updated on every make.

3564. By Jon Hill on 2011-08-15

Bug fix to Momentum_DG when using implicity_buoyancy. The interface to ele_grad_at_quad got changed a while back and this one was missed or reverted

3563. By Rhodri Nelson on 2011-08-15

Switching the indices of an array which seems to have been missed when the ele_grad_at_quad interface was updated a while back.

3562. By Patrick Farrell on 2011-08-15

I think this will fix the cx1 build. The rule to fetch the python
binding to libspud (libspud.so) was trying to copy it from
lib/python*/site-packages, but redhat helpfully puts the python binding in
lib64/python*/site-packages.

3561. By Patrick Farrell on 2011-08-15

This branch contains the features that let the embedded python interpreter in fluidity access the spud dictionary of fluidity's options.

Use it as follows:

import libspud
libspud.get_option("/simulation_name")

It only works for Python >= 2.7, so if you want your code to work on 2.6 or before, you need to guard with an

if sys.version_info[1] > 6

This branch also contains some minor fixes for compiling fluidity against; sometimes things were using the system one, and sometimes things were using the libspud/ directory.

This branch also updates libspud/ to be the latest spud trunk. Those changes have been reviewed earlier in other merge requests.

A lot of the work on this was done by Nan Mao.

3560. By Patrick Farrell on 2011-08-15

Put in a dependency on include/version.h. This fixes a race condition
exposed with make -j since the recent change to how the version number
is handled.

3559. By Jon Hill on 2011-08-13

Bug fix in vector_field_addto_field_scale_field. If we remap scale then we should use the remapped field, not the original

3558. By Lawrence Mitchell on 2011-08-11

buildsystem: Move version definition out of include/confdefs.h

Setting the version number of Fluidity during configure is the wrong
thing to do, and writing it into include/confdefs.h doubly so.
Instead, add a new rule to the top-level Makefile to generate
include/version.h. If you need to know __FLUIDITY_VERSION__ you must
now explicitly include version.h as well confdefs.h.

While we're at it, remove any last vestiges of svnversion from the
configure scripts.

3557. By Tim Bond on 2011-08-12

Description of the Change from Patrick:

  Extensions to the python-state system to expose CSR matrices.

  User code can access CSR matrices like:

  m = state.csr_matrices['PressureMeshMassMatrix']

  The interface is quite extensive, as it subclasses scipy.sparse.csr_matrix. So
  you can do almost anything with it: multiply it by a vector (m * field.val),
  solve mu = b (scipy.sparse.linalg.spsolve), etc.

  This code was written by Nan Mao; I just cleaned it up and prepared it for
  merging.

  Note that it only exposes CSR matrices that happened to exist in state when it
  was passed to calculate_diagnostic_fields. In many cases, that appears to be
  precisely 0. One can test this by adding a call to

  get_mass_matrix(state, velocity_mesh)

  just before the call to calculate_diagnostic_variables. When this is merged, I
  will investigate the lifecycle of the CSR matrices in state so that there is
  actually data to expose.

3556. By Stephan Kramer on 2011-08-11

Bug-fix in interpolation manager for periodic meshes. This fixes bug https://bugs.launchpad.net/fluidity/+bug/822355 reported by Flora.