Closed arporter closed 6 years ago
The logical conclusion is that the problem is something to do with the new support for handling comments, even though I thought that was just for fparser2. Maybe I modified the reader too?
Yes, it seems I changed the default behaviour regarding comments - they are now ignored by default. This means we have to edit CommentGen
and DirectiveGen
in f2pygen.py to tell the readers that they create not to ignore comments.
Once we do that we'll get failures on Travis as that tests against the latest release of fparser (which doesn't yet have this feature in it). However, we already face this problem in #185 as that makes use of the new Comment support in fparser2.
After discussion with @rupertford, we have elected to solve this problem by making fparser a git submodule of PSyclone (with the proviso that when we make a release it must work with a released version of fparser). That work is being done in #212 so I'm closing this issue.
Last commit to fparser that works was: 92c4c21d70ccebcc28ce4133dede187fa7769371:
The (fparser) changelog since that merge is:
08/08/2018 PR #96, for issue #95. Removes dependency on nose and tidy test_fortran2003.py for pylint.
01/08/2018 PR #94 for issue #92. Re-structures fparser2 and introduces a parser factory. This creates a parser for a specified Fortran dialect (currently 2003 or 2008). Intoduces support for Fortran2008 submodules.
09/07/2018 PR #90 for issue #89. Make fparser2 pycodestyle conformant.
09/07/2018 PR #88 for issue #81. Bug fix for undefined variable in fparser class 'HasImplicitStmt'.
07/07/2018 PR #71 for issue #68. Adds support for keeping input comments in the Fortran output for parser 2.
27/06/2018 PR #85 for issue #82. Adds support for the full list of possible procedure attributes (POINTER and PROTECTED were missing).