Open tlienart opened 3 years ago
I completely agree on disabling minification by default! :smile: Would make debugging problems (from Slack) much easier in many cases.
I totally missed this issue! But see my discussion here: https://github.com/tlienart/Franklin.jl/discussions/927 TLDR I think the current minifier should not be used, since the performance gain is tiny (because of gzip) but the headache cost is large (partly because the project is not well-tested and no longer maintained).
My ideas are could remove it completely, or to use a modern bundler like parcel.
Thanks @fonsp I'll answer to your careful suggestions in the thread, my current inclination is to just remove it and let users do their own minification if they want to (initially I had this naive idea that minification was useful + easy and now I know better...)
Great! It would be good to leave the minify
kwarg for backwards compatibility, and just ignore the value if false
, and print a message when true
.
There's too many corner cases:
tbody
disappearing into the void)In an ideal world both of these things are nice, but it'll be a better experience if users opt-in rather then the opposite (and then we can ping them about testing stuff before they deploy).
This is effectively a breaking change, this should be changed after users are encouraged to pin minor releases of Franklin to prevent issues.