The problem is caused by the fact that we have no mechanism of specifying CSS dependency order in Royale.
The design of the CSS makes sense. Requiring a specific order for the CSS files is a valid strategy. NPM (via modules) does have a way to specify CSS order. I think this is a general problem which we should try to solve. To me, the only question is technically what’s the best way to do so. Possibly inject_html is the wrong tool (as we’ve already discussed), but it’s the only one we have as of now. I’m not sure how the order of the inject_html is determined. Maybe there should be some sorting logic which corresponds to dependency trees? Maybe the right answer is to have a way to declare file dependencies and that should include order as well? Harbs > On Feb 12, 2020, at 1:54 PM, Carlos Rovira <[email protected]> wrote: > > Hi Harbs, > > if I understand the issue correctly this not seems a Royale issue to me, > but a problem in Spectrum CSS structure design. > In Royale, we provide CSSs to the compiler from libraries, and that > generate just one compiled CSS file. User's project CSS takes precedence > over framework libraries, so if a user override a style, that wins over the > royale one. Since load an external CSS is outside the Royale sandbox, I > think the loaded CSS should solve its problem to not have > collision problems within them. > > > El mié., 12 feb. 2020 a las 10:04, Harbs (<[email protected]>) escribió: > >> I’ve run into an interesting issue while working on the Spectrum >> components.[1] >> >> To quickly summarize: >> >> I’m using inject_html to automatically add CSS links to the index.html >> file. Doing so gives no control over the order of said dependencies. >> >> The CSS files are required to be loaded in the declared dependency list >> (in the npm package). >> >> I think we need to resolve this issue somehow and I’m not sure of the best >> way to go about it. >> >> Thoughts? >> Harbs >> >> [1]https://github.com/adobe/spectrum-css/issues/481 < >> https://github.com/adobe/spectrum-css/issues/481> > > > > -- > Carlos Rovira > http://about.me/carlosrovira
