aboutsummaryrefslogtreecommitdiff
path: root/runtime/ftplugin/javascriptreact.vim
diff options
context:
space:
mode:
authorJames McCoy <jamessan@jamessan.com>2019-11-13 22:20:57 -0500
committerJames McCoy <jamessan@jamessan.com>2019-11-13 22:31:02 -0500
commit4c48cf3020202a614d826f2ef2893380a3e9c8cc (patch)
treed86d47c3aacbe895132319bfcf8369db7f80d174 /runtime/ftplugin/javascriptreact.vim
parente2cc5fe09d98ce1ccaaa666a835c896805ccc196 (diff)
downloadrneovim-4c48cf3020202a614d826f2ef2893380a3e9c8cc.tar.gz
rneovim-4c48cf3020202a614d826f2ef2893380a3e9c8cc.tar.bz2
rneovim-4c48cf3020202a614d826f2ef2893380a3e9c8cc.zip
f_getenv/setenv: Access v_special when v_type is VAR_SPECIAL
Multiple Debian builds were failing these tests: Failures: From test_environ.vim: Found errors in Test_external_env(): function RunTheTest[37]..Test_external_env line 16: Expected '' but got 'FOO=null\n' Found errors in Test_getenv(): function RunTheTest[37]..Test_getenv line 2: Expected v:null but got v:false Found errors in Test_setenv(): function RunTheTest[37]..Test_setenv line 5: Expected v:null but got 'null' This is because nvim has a separate tag (`v_special`) in `typval_T` for special variables, whereas vim re-uses the `v_number` tag. On little-endian architectures, using the incorrect tag is not an issue because the byte representation is the same. However, on big-endian systems this caused the `v_number == kSpecialVarNull` checks to fail, and the non-special code to execute.
Diffstat (limited to 'runtime/ftplugin/javascriptreact.vim')
0 files changed, 0 insertions, 0 deletions