We dont want to set the loc list by default since syntastic isnt the
only plugin using loclists - and we dont want to clobber other peoples
loclists.
This option allows the user to override this behaviour if it suits their
workflow.
JSHint has a secondary error format that includes a number related to the error message.
E.g. The normal error format is something like:
.\Foo.js: line 2, col 5, Missing semicolon. (W033)
But the secondary error format includes a number in parentheses:
.\Foo.js: line 3, col 4, Blocks are nested too deeply. (2) (W073)"
I've changed the errorformat to read the message until it hits a space, instead of a period, so that these errors are included.
This reverts commit 0ef513a0a8f14ad8194a3300ec2563761a1334fe.
The cases that I was trying to catch here were already caught further
along in the errorformat
pass `jshint` the `--verbose` option to get 'W' or 'E' indicating
whether something is a warning or an error.
NOTE: this was tested with jshint 1.1.0 installed with `npm install
jshint`
It now uses the correct path for the dart_analyzer, it uses the
machine readable format to provide more information.
The highlighter now highlights the exact span that the analyzer complains
about. Further, warnings are different from errors now.
Seems that `:sign define ... texthl=bar` causes `hlexists('bar')` to
return true.
So set up the highlight groups before the signs - otherwise they will
never be set up by default.
Previously we removed the code to automatically set the loclist and only
set it when required. This was needed because otherwise we could
interfere with other plugins that use the loclist.
Add the call to setloclist back in only if syntastic_auto_jump is set.
This will blow away the loclist every time the file is checked, but
setting syntastic_auto_jump is the users choice.
Fixes#533
The variable force_highlight_callback is gone.
Highlight functions are now consistently named
SyntaxCheckers_<filetype>_<checker>_GetHighlightRegex(), and they
take precedence over highlighting based on column.