Daniel Alley

@Mingun (and @tafia), what are your thoughts about allowing nominally specification-breaking, but widespread behavior like this? Is this something that might be acceptable under a feature flag?

Let's open an issue and collect a bit more information before we merge anything - for instance, what do other XML libraries do. It looks like [nokogiri](https://github.com/sparklemotion/nokogiri/issues/1429) may have at...

It appears that C#'s dotnet XML library has a flag for this: https://learn.microsoft.com/en-us/dotnet/api/system.xml.xmlreadersettings.checkcharacters?view=net-6.0 Same for writing: https://learn.microsoft.com/en-us/dotnet/api/system.xml.xmlwritersettings.checkcharacters?view=net-6.0 >Surprisingly, the only restricted character which is actually restricted is the escaped NULL:...

> The following behaviors are possible: > > * emit hard error (= stop parsing / writing and return error) > * silently skip invalid characters (or at least log...

Does that sound right @sashka ?

@sashka Do you plan to continue with this?

HTML predefined entity resolution is a decent representative benchmark for this, I think. With this one function enabled, compilation takes 30 seconds, compared to 3 seconds without it. https://github.com/tafia/quick-xml/issues/763

@JOSEPHGILBY Do you intend to continue this work?

Hey, so I'm writing a library that handles parsing some compressed XML files and noticed that `miniz_oxide` is responsible for the great majority of the runtime of my testcase program....

I just recompiled my application w/ the flate2 "cloudflare_zlib" feature instead of the miniz_oxide backend, and it knocked nearly 30% off of the runtime. Possibly a good place to mine...