Files with same name can produce same css classes

[mlmorg]

If you have two files with the same name (e.g. foo.css) both with the same local css class (e.g. .bar) but with different declarations, the resulting css looks something like:

._foo__bar {font-weight: bold;}
._foo__bar {font-weight: normal;}

What are your thoughts on the best way to handle this?

[rtsao]

The Webpack implementation allows you appends a hash of the content to the class name.
https://github.com/webpack/css-loader/blob/master/lib/getLocalIdent.js#L13
https://github.com/webpack/loader-utils/blob/master/index.js#L226

We need to be able to configure https://github.com/css-modules/postcss-modules-scope/blob/master/src/index.js#L159 to do something similar, but currently only the name and path are passed as arguments. Otherwise, we should be able to override generateScopedName through the options object.

[joshwnj]

Good call. I think it would be good to look into matching the behaviour of the webpack implementation but at this stage not sure how deep that goes - it may be a lot of work for not much value. We may well find some other method of deduping that fits easier with the workflow.

[geelen]

This is strange, I thought the name being generated is the full path of the file, so it would be ._lib_foo__bar and ._util_foo__bar. That's how this line is intended to work: https://github.com/css-modules/postcss-modules-scope/blob/master/src/index.js#L160

[joshwnj]

@geelen yeah that's what I thought too... but it seems under certain circumstances it won't work that way. I'd like to try to get full paths working as well, because it gives some useful info to the developer. But I'm thinking it would be good to be able to use the hash for uniquifying as well (since that's what webpack is doing too).

[geelen]

If you're just hashing the path & the classname, won't the hashes be identical too?

[joshwnj]

nah I'm hashing the source css

[geelen]

Ah cool, alright i'll push through your change this arvo

[joshwnj]

cheers!

[geelen]

postcss-modules-scope@0.0.8 released, guessing this is now covered under #16 so I'll close this

[mlmorg]

Not quite. css-modules-loader-core needs to have postcss-modules-scope bumped to 0.0.8.

Better yet, bump everything to 1.0.0 so that you don't need to do those bumps manually anymore :)