diff options
| author | Nathan Sidwell <nathan@acm.org> | 2020-07-21 12:37:58 -0700 |
|---|---|---|
| committer | Nathan Sidwell <nathan@acm.org> | 2020-07-21 12:42:09 -0700 |
| commit | 4d6e94960aa199b2fe7bf5a66f98bd7691884c1d (patch) | |
| tree | a9e4bce1eb8d4b9514d0baa8322d95d61fb1e0fa /gcc/fortran/parse.c | |
| parent | 28f2a080cc27531a8c78aec9f44aeff4961c2a4c (diff) | |
| download | gcc-4d6e94960aa199b2fe7bf5a66f98bd7691884c1d.zip gcc-4d6e94960aa199b2fe7bf5a66f98bd7691884c1d.tar.gz gcc-4d6e94960aa199b2fe7bf5a66f98bd7691884c1d.tar.bz2 | |
c++: Fix scan forward over pragma [PR96257]
It turns out that the paren scanning code is used for speculatively searching
to see if we're looking at a compound_literal. So we shouldn't always purge
pragma tokens.
gcc/cp/
* parser.c (cp_lexer_consume_token): Drop PRAGMA_EOL assert.
(cp_parser_skip_to_closing_parenthesis_1): Only pass start token
to pragma skipper if recovering.
(cp_parser_skip_to_pragma_eol): Only purge and change pragma
state when recovering.
gcc/testsuite/
* g++.dg/parse/pr96257.C: New.
Diffstat (limited to 'gcc/fortran/parse.c')
0 files changed, 0 insertions, 0 deletions