On Monday, 12 February 2018 at 21:51:56 UTC, H. S. Teoh wrote:
We can even design the DTD support wrapper to start with being
just a thin wrapper around dxml, and lazily switch to full DTD
mode only if a DTD section is encountered. Then user code that
doesn't care to use dxml's raw API won't even need to care
about the difference.
In this vein, if a new version of std.xml didn't offer pure and
fast parsing like dxml, but included DTD by default, people would
complain that that was the real deal breaker (too slow, man!).
Remember `autodecode`? Right.
DTD inclusion should only be available on demand. Imagine you
want to implement a library project where ebooks (say classics)
are catalogued and presented in an ebook reader on the web (or in
an app on your smart phone). It is likely that the whole DTD
thing would probably be done at the cataloguing stage, but once
the books are in the library most users will probably just want
to go through them page by page or search for quotes etc. - and
for that you'd need a fast tool like dxml with no overhead.