[gsnedders]

epubcheck is meant to ensure conformance with the standards, not the interoperably implemented subset of the standards. (Which has lots of awkward questions: which implementations of the standards, which versions of those implementations, etc.)

[ameliaquining]

The latter seems like what the tool's users actually want. That it's a harder problem doesn't change that.

[MadnessASAP]

The user wants the website to work in IE6, developing and testing only against IE6 to the detriment of other browsers is not generally regarded as a healthy state of affairs.

The standard exists, it is the responsibility of both the producer and consumer of ePUB files to adhere to the standard.

https://en.wikipedia.org/wiki/Robustness_principle

[ameliaquining]

If a large fraction of your users are on IE6 and you can't realistically get them off it, you need to make sure your site works in IE6, and good tooling should help you do this. Of course you also want to make sure it works in other browsers your users use, and a standard may be helpful in doing that.

[Finnucane]

No, the users need to be able to check for conformance. What we also need is for vendors to supply test platforms. Amazon, to its small credit, does this, which is good, because the subset of html/css they support is limited and poorly documented. Heck, I'd be happy if Apple, Kobo, and everyone else just kept good documentation and up to date!

Though these days I have to spend more time worrying about EAA and ADA compliance than anything else.

[ameliaquining]

A compatibility linter is a poor substitute for a vendor-supplied test platform, but if the vendor is uncooperative it may be the best that can be done.

[Finnucane]

It's not a direct substitute at all. It's not intended to be. And--it's on the vendors for making crap software and not keeping up.