stylus / stylus Goto Github PK

my bad!

+ / > do not need to retain whitespace, start compression tests

move most to separate markdown files, we can use these for the site and the readme.

for right now it helps with debugging since otherwise every id might be a variable. This is gutsy, but if it works, rad

some are randomly off, sometimes perfect

break out into a util and unit test it

http://reference.sitepoint.com/css/at-fontface

for example currently 5s == 5ms is true, which.. is not true :p

ident might become the node for variables, selectors, and properties. However when the context is a selector we will need to stitch idents together to form the selector. worth looking into

For example 14px/1.5 needs to either be escaped, or we can only perform arithmetic when parenthesized (that is lame), so I guess maybe 15px\/1.5 ?

• global from a mixin
• local from nested mixin
• mixin within a mixin
• function within a mixin within a mixin

all that shit

separate selector nodes, tests etc. Make sure that recursion is applied to all of them

throws

tries to show context of the importing file

test coverage as well for many of the cases. stuff like an error in an @import script will try to show you the parent script instead

might already be working, will have to check it out

@fn = red
@fn(#fc0)
 // => 255

no tests for nested yet

with registerExtension() trick

test for mixins and functions

not a huge fan, but there are times when it would be nice to have:

some selector ${@var}
  color red

etc

recursively apply

currently for example hsla() called with only 3 args will result in "Cannot read property 'constructor' of undefined", we need better errors here, quick fix

one issue is a function return val vs mixin:

foo(n)
  some-prop -n

"looks" like a prop with a negated n to us, but to the parser we will need some magic, possibly even lazy evaluation to treat this as a mixin or a function call, or require that return is used (hopefully not)

we could also have a slightly different syntax for functions, which is kinda lame too, although you most likely will never have a function act as a mixin and have a retval

for all the jQuery people out there pad ( 5px ) instead of the logical pad(5px)

transparently via make stylus.js and have make stylus.min.js. This will be a reasonable hack but might work with some require() replacement etc. I would like to keep this out of the source at least, browser compat is uglayyyYYY

I think one of the only "real" applications for this is a demo editor on the site

these seem really annoying to create lol painful grammar definition but would be really awesome to have, same with vim and friends

currently created like an unary not on an ident

we have to ensure proper stitching, so we have to track if trailing space was present or not

from a scanning stand-point its invalid to have improperly indented lines regardless of them being "blank", but some Lexer logic can avoid these

variable mixins / functions are working fine but not BIFs

root level mixins should be possible as well

I guess less.js fucks up on modern stuff there, I havent even tried, so add tests and implement

typical stuff

if / else / unless blah blah, another layer of mind-bending context madness

tough issue due to all the ambiguity going on, but yeah.. fix me

omission of the extension will attempt to inline the AST from the given path, otherwise *.css is literal. Finish up

https://github.com/NV/CSSOM

something like:

stylus --convert < in.css > out.styl

or

stylus --convert *.css

im a big fan of simple std streams but this might be more convenient

it is nice to abstract, but might be worth adding them as methods instead for BIFs and extensions

it is not very good, and could be done with a tiny parser, since it does not support @font-face etc

Stack#lookup() breaks with nested selectors in a mixin

currently the error reporting lineno is based on the last scanned line, which works fine until you start calling functions. Need to store the lineno and reference it on Call