Mailing List Archive

Support open source code!


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [tlug] Selective mojibake email



On Fri, Mar 01, 2002 at 04:39:00PM +0900, Stephen J. Turnbull wrote:
> 
>     >> The fact is that Microsoft does not properly implement
>     >> standards that it doesn't control, nor even its own.
> 
>     Matt> Steve, that's unfair.
> 
> I disagree.  I think it's accurate.

Hmm. First of all, I apologize for the nasty edge to my previous
remarks. I don't know why I found it necessary to attack like that.
Maybe the combined stress of trying to survive in the crumbling US
economy, worrying about what kind of mess our illustrious leader
will get us into next, and raising a 2-year old is starting to get
to me.

>     Matt> If you said something like "usually does not" or "does not
>     Matt> properly implement i18n-related standards,"
> 
> By "does not implement" I meant "I know of no situation where
> Microsoft will acknowledge non-conformance as a bug, until they
> confirm that non-conformance will cost them market share."

Fair enough.

>     Matt> But at least in that field, their management finds it
>     Matt> expedient for the time being to play by the rules;
> 
> And they will not hesitate to break existing functionality that
> conforms, should the fit take them.  In fact, they regularly do this
> strategically, with malice aforethought.

Agreed.

> Standards are useful not because they tell you what to do when it
> doesn't matter.  Standards are useful because they tell you what to do
> when you desperately want to do something else.  If you conform in the
> former case and not the latter, you are not implementing the standard.
> 
> Microsoft's XSLT conformance is an accident.

That depends on how you define "accident." If you mean something like
a historical accident, in that the corporation accidentally found itself
in a position where it seemed best to comply with the standards, I'll
let you get away with that one. But the record, for those who care to
take a look, shows that Microsoft started out being very non-standard;
the developer community complained; Microsoft listened, and brought
their product into compliance. Sounds pretty deliberate to me.

I also spent a year working for one of Microsoft's favorite XML spokes-
men. During that time part of my job was delivering XSLT training to
Microsoft employees. I've met the lead developers of their XML parser.
What I learned from that experience was that Microsoft isn't a nest of
demons. There are many good, intelligent people working there who, as I
said, are sincerely trying to do the right things. The problem is just
that it's a large company with too much power and a warped sense of 
values. Seems like there's more than one of those around. 

So, do I trust Microsoft? Uh-uh. I was just trying to point out that
their XML processor is, if not 100% compliant with the spec, very,
very, close -- at least as close as the IBM, Oracle, and Apache products.
It's also, if I remember right, SAX- and DOM-compliant. And it's very
fast. And it offers some very useful extensions that make developer's
lives easier (though they don't make much effort to point out that
those extensions *are* extensions -- for which I have publicly criti-
cized them). Now I have a feeling you would say that their compliance
in this one area is, in the grander scheme, intended to support a
strategy that ultimately goes against standards. Well, I wouldn't be
surprised. But a good product is a good product.

  And if you depend on it
> to continue, you are an accident waiting to happen.

Not me. I had to put in my time in a Microsoft shop to break into the
XML biz, but thankfully I'm out of that. Recently, I have been working
with Fourthought, which is pretty close to ideal. All Python all the
time (except when clients make us use GASP, COUGH, RETCH, Java :), and
with a strong pro-Linux bias. Of course, our products have their bugs,
(more than MSXML, I'm afraid) but at least we admit it.

> In comp.lang.python <3C7BB093.3020600@example.com> (excellent thread, see

Thanks. I'll check it out. 

-- 
Matt Gushee
Englewood, Colorado, USA
mgushee@example.com
http://www.havenrock.com/


Home | Main Index | Thread Index

Home Page Mailing List Linux and Japan TLUG Members Links