~asc/fluidity/regularisespatialaspect : changes

~asc/fluidity/regularisespatialaspect » Changes from revision 3530

From Revision 3512 to 3493

Rev	Summary	Authors	Date
3512	femtools: Fix r3511 for old PETSc versions femtools: Fix r3511 for old PETSc versions IsParallel is not pure, and therefore we can't call it from a pure function. So remove the pure attribute from petsc_must_assemble_by_column.	Lawrence Mitchell	12 years ago
3511	femtools: Pull check for column-wise assembly into separate ... femtools: Pull check for column-wise assembly into separate function A bug in PETSc (prior to version 3.0.0-p8) requires that PETSc matrices with MAT_IGNORE_ZERO_ENTRIES set are assembled column-wise. This work-around is not required for newer releases. Introduce a new function which only does the check if the PETSc version is old enough to have the bug (pre 3.0.0-p8). This should hopefully be inlined appropriately, speeding up matrix assembly in some cases.	Lawrence Mitchell	12 years ago
3510	Don't use hashlib module in make_{check_options,register_dia... Don't use hashlib module in make_{check_options,register_diagnostics} The hashlib module only appeared in Python 2.5, superseding the now deprecated sha module. Rather than requiring a new Python, just use sha directly (the results are the same).	Lawrence Mitchell	12 years ago
3509	Reduce spurious recompilations in clean or mostly-clean tree... Reduce spurious recompilations in clean or mostly-clean trees Avoid touching lib/libfluidity.a unless actually necessary. This comes in a few parts: o Use bzr revision-info to set up Fluidity's version number (and only update confdefs.h if the version changed) o Calculate hashes of the check_options and register_diagnostics files and only update them if the hash changes o Avoid depending on .PHONY targets that just point directly at actual targets	Lawrence Mitchell	12 years ago
3508	Trivial fix for not really a bug (unless we start requiring ... Trivial fix for not really a bug (unless we start requiring that special tests run).	Cian Wilson	12 years ago
3507	Unperiodising a periodic mesh in parallel may produce floati... Unperiodising a periodic mesh in parallel may produce floating nodes that are unattached to any element on that process. This leads to linked lists with zero length in make_sparsity (for the unperiodised mesh) resulting in a segfault when requesting the maxval. This commit prevents this from happening by not checking for the possible number of columns associated with floating nodes - i.e. that have a 0 length linked list.	Cian Wilson	12 years ago
3506	Removing one of the two readme files in the legacy_reservoir... Removing one of the two readme files in the legacy_reservoir_prototype dir to avoid confusion in non-case-sensitive environments	Jon Saunders	12 years ago
3505	Commit in response to bug 802426: Commit in response to bug 802426: https://bugs.launchpad.net/fluidity/+bug/802426 This only affects the build process and is thus orthogonal to the code base. Summary: On RHEL 5.6 with GCC 4.4 a requirement for -llapack to occur after -lmba2d was noted, causing a build failure. This is remedied by this commit by swapping the order of '$LIBS -lmba2d' in configure.in.	Timothy Bond	12 years ago
3504	Bug fix for the bug fix in r3503. Thanks to Stephan for spo... Bug fix for the bug fix in r3503. Thanks to Stephan for spotting this. Shouldn't affect the tests that use this as they already have dimensions of order 1.	Cian Wilson	12 years ago
3503	In response to Tim's request... the test is failing on build... In response to Tim's request... the test is failing on buildbot 32 bit because of precision issues in numpy lexsort. Fixing this by rounding things less than (in an absolute sense) machine precision to zero before calling lexsort in VtuMatchLocationsArbitrary. I don't have a 32 bit machine to test this on so will keep an eye on buildbot. All tests that use this function pass locally.	Cian Wilson	12 years ago
3502	make plot script consistent what is used for the wetting and... make plot script consistent what is used for the wetting and drying paper	Simon Funke	12 years ago
3501	An interface from libadjoint (adj_create_variable) has chang... An interface from libadjoint (adj_create_variable) has changed to be more Fortran-ish. This change updates the adjoint code to match that change in the interface. If this doesn't compile on buildbot, just wait a day until the new libadjoint has been built and pushed out to the builder.	Patrick Farrell	12 years ago
3500	Moving the test case from 4 to 2 processors so hopefully it ... Moving the test case from 4 to 2 processors so hopefully it will play nicely with zoltan	Hannah Hiester	12 years ago
3499	Updating the test tolerances with a slight slackening to res... Updating the test tolerances with a slight slackening to respond to what looks to be a small OS-level change.	Timothy Bond	12 years ago
3498	fix two or the radiation unit tests that were failing for gf... fix two or the radiation unit tests that were failing for gfortran 4.6 for the first 4.6 didnt like reading lines beyond the length of the file, which was a mistake, and the intented action wasnt actually required for the test so removed. for the second a 1d array was allocated length 1, but then by mistake values set as (/1,11/) which for 4.6 I presume implied a dealloc re-alloc automatically which then caused the test to fail when the size of the array was checked	Brendan Tollit	12 years ago
3497	Fix bashism in test. Please don't use &> in tests. Fix bashism in test. Please don't use &> in tests.	Stephan Kramer	13 years ago
3496	Remove incorrect tag on this test Remove incorrect tag on this test	Jon Hill	13 years ago
3495	Colin forgot to commit this I think. Colin forgot to commit this I think.	Stephan Kramer	13 years ago
3494	Need to use $(FC) not $(LINKER) Need to use $(FC) not $(LINKER)	Colin Cotter	13 years ago
3493	Fixing a minor bug in create_makefile which was causing the ... Fixing a minor bug in create_makefile which was causing the special-casing of mba2d to fail.	david.ham at ac	13 years ago

Older »