[
Lists Home |
Date Index |
Thread Index
]
Just as a conlcusion of the thread here is the rest of the communication
(with consent of Derek), which did run past the list. I believe that this
gives a definate answer and that might help others using XmlTextReader. It
also gives some outlook into a (hopefully) brighter future.
-----Original Message-----
From: Derek Denny-Brown [mailto:derekdb@microsoft.com]
Sent: Donnerstag, 10. Juli 2003 17:48
To: Eckenberger Axel
Subject: RE: [xml-dev] Handling of significant whitespace in .NET XmlReader
> > True. The decision was made to deviate from the letter of the
> > standard because it was deemed to be in the best interest of the
> > majority of the customers. The information wasn't lost, so other
> > customers were not being ruled out.
>
> I grant you that about 70% - 80% of the xml being processed is data
> centric so that's design decision is OK. The problem is that you have
> to
specify
> the
> where you deviate from the standard, otherwise developers, like
myself,
> are
> lead to believe that a given XML parser parses XML without altering
the
> content - which is a fair enough assupmtion if the parser claims to
> conform to the XML standard.
<rant style="anti-large-company">
Yes, but you have no idea how hard that can be to actually do at a large
company. The mere fact that XmlTextReader is _so_ non-conformant by
default, (and enabling conformance makes is much slower) was something I
strongly argued against. I lost. </rant>
> Well that's allright if you start up a project, but we are faily far
down
> the road ...
>
> At the start of the project we decided to use .NET (and this at a
stage
> when
> it just had come out! ... brave decision ... ;-) ) because of its
> andvantages (and as the logical successor to COM which was defined by
our
> customer), so open source projects were not really available to us or
only
> at the high cost of cross language implementations.
>
> Hoewer, the problem with the parser arose because of a non-documented
> non-conforancy that provided a pitfall in which we have promply
fallen. If
> the problem were discovered early on or we had known about, we could
have
> saved some precious development time, which in the current situation
isn't
> freely available anymore.
I can say, that the next major release will address much of your concerns.
There will be a faster, much more conformant parser, and we will be docing
XmlTextReader as the less preferred option.
> So what I (as a customer and developer using MS products) would whish
for,
> is that you (Microsoft) would provide me with all the necessary
> information to make an educated desing decision early on in the
> project and thus provide
> the pitfalls you can step into.
I will mention your frustrations. It is good (to me at least) to hear some
solid word-from-the-trenches about why standards conformance is so
important. It helps me argue for better conformance, and better understand
how to mitigate our decisions to choice non-conformant behaviour in some
cases.
Most excellent to hear that you are using our frameworks, and if you have
other issues, feel free to email me directly.
Cheers
-derek
|
