Scott Allan wrote:
> At 10:51 PM 9/9/01 -0700, Gregory Neil Shapiro wrote:
> >sallan> Is it not possible to easily script against the RWI with most dev
> >sallan> tools these days? I know I have seen stuff floating around that
> >sallan> does this...
> >
> >For some activities, there are hacks floating around which try to
interpret
> >the HTML that is returned. Hacks like that tend to be extremely fragile.
> >This is far cry from a clean API.
>
> Fair. I suppose my understanding of the issue was not clear... that's why
I
> asked. :)
Let me second what Gregory said here and help perhaps help make things
crystal clear -- the RWI scripts do tend to be fragile and inefficient. I've
written a number of them myself and I always cringe at the programming
techniques I have to take.
The problem is that they are extremely dependent on the un-published
specification of the HTML tables and layout and encoding of values in the
RWI. There is no specification because the RWI is designed for human use and
all the fields are easily human-understandable. For example, it is entirely
reasonable for OpenSRS to change without warning the display of an "order
approved" value from no/yes to approved/denied. Humans would easily
understand the change and probably not even notice, but a script would
choke.
An API designed for computer consumption would clearly define these
parameters and the programmers (on both ends) would stick to the
specification, thus preventing problems.
Scripts working through the RWI also have the drawback that they cause more
server and network load than a clean API would. Especially back in the day
when people downloaded their entire domain listings a page at a time.
David
