1. Bug reports / GitHub issues
Please note that the preferred channel for posting bug reports is the issue tracker at GitHub. Reports posted elsewhere are less likely to be seen by the core team.
When reporting a bug make sure you search the existing GitHub issues
for the same/similar issues. If you find one, feel free to add a
comment with any additional information that may help us solve the
When creating a new issue be sure to state the following:
- steps to reproduce the bug;
- the version of Vim you are using (run
:verto find out);
- the version of syntastic you are using (see
For syntax checker bugs also state the version of the checker executable that you are using. Adding debugging information is typically useful too:
- open a file handled by your checker;
g:syntastic_debugto 1 or 3;
- run the checker;
- copy the output of
2. Submitting a patch
Before you consider adding features to syntastic, please spend a few minutes (re-)reading the latest version of the manual. Syntastic is changing rapidly at times, and it's possible that some features you want to add exist already.
To submit a patch:
- fork the repo on GitHub;
- make a topic branch and start hacking;
- submit a pull request based off your topic branch.
Small, focused patches are preferred.
Large changes to the code should be discussed with the core team first. Create an issue and explain your plan and see what we say.
Also, make sure to update the manual whenever applicable. Nobody can use features that aren't documented.
3. General style notes
Follow the coding conventions/styles used in the syntastic core:
- use 4 space indents;
- don't use abbreviated keywords - e.g. use
endfun(there's always room for more fun!);
- don't use
l:prefixes for variables unless actually required (i.e. almost never);
- code for maintainability; we would rather a function be a couple of lines longer and have (for example) some explaining variables to aid readability.
4. Syntax checker notes
Make sure to read the guide if you plan to add new syntax checkers.
Use the existing checkers as templates, rather than writing everything from scratch.
The preferred style for error format strings is one "clause" per line.
E.g. (from the
let errorformat = \ '%E%f:%l:%c: %trror: %m,' . \ 'Syntax%trror: In %f\, %m on line %l,' . \ '%EError: In %f\, Parse error on line %l: %m,' . \ '%EError: In %f\, %m on line %l,' . \ '%W%f(%l): lint warning: %m,' . \ '%W%f(%l): warning: %m,' . \ '%E%f(%l): SyntaxError: %m,' . \ '%-Z%p^,' . \ '%-G%.%#'