[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: well-formedness error
At 10:23 04/06/19 -0400, James Robertson wrote:
I'll point out again that I handle bad XML (not missing tags, but a lot of
other errors) in BottomFeeder - and I have not
-- written a custom parser
-- do not use Regex
-- do not use the Universal Parser
I spent an aggregate total of maybe 15 minutes on this entire problem,
broken up into a handful of 3-5 minute sessions over the course of a few
months.....
Okay, so that's how it starts. That's probably similar to how it started
at Netscape, Microsoft, and a few other 'browser vendors'. But that's
just the start.
What happens next is that some users tell some other agregator (or similar)
software creator that their feeds work on BottomFeeder, but not on that other
agregator. The programmer of that agregator spends a few hours to try to
tweak his parser. After all, customers are important. Because he uses
other tools, and is otherwise not aware of exactly what you were doing,
what he parses is slightly different from what you parse.
So next, some users come to you and tell you that some feeds parse on
that other agregator, but not on BottomFeeder. So next you spend a few
days to figure out what you have to tweak in your parser to accept that
feed.
You can easily see where this leads. What started with a little 15-minutes
curtesy to some lazy users turned into a major headache for a whole industry.
And this is not some fantasy story that I just made up, this is more or less
what happened for html. Nobody who has been there ever wants to go there
again. Some companies clearly have learned their lessons. Why do you think
Microsoft has spent that much time on XML, and continues to spend time
on XML to find even the lates little unclarity in the XML spec? And
why do you think they managed to make sure that a document that is
not well-formed produces an error? (as others have said, there are
some bugs, and there are some shortcuts for performance, but they
don't come anywhere near to any bugwards-compatibility slippery
slope)
If you want to go there, that's your choice. If you can convince
the Atom community to follow you, that's their choice. But if
that happens, please make sure that never ever in your spec
you use the word "XML".
Regards, Martin.