> I understand where Chris Love is coming from.  I think OpenSRS should,
> first and foremost, provide a well-documented protocol to their service.
> Then they should provide an implementation.

Actually, I was able to build the PHP client using nothing more than the
published documentation and the Perl classes as a starting point.  So I
think their documentation is sufficient.  It's not "complete", or
"well-done" by any means, but there is enough there. :)

As far as the encryption goes, part of the problem I had in porting the code
to PHP had nothing to do with DES or Blowfish, but had everything to do with
the CBC mode in Perl's Crypt::CBC.  If anything is non-standard, this would
be it.


> My biggest concern isn't the initial barrier to figuring things out, but
> rather when small changes on their side become mountains on my side.
> So, if it takes me 1 man week of work to write a java client, that's fine.
> But when each new feature takes an additional man week to solve, then that
> becomes a serious issue.  Should that time ever occur, I believe we would
> seriously consider a new registrar.

However, once you have a working client that can encrypt, decrypt, read and
write the proper XML messages, you pretty much have no work left to do.
Every new feature OpenSRS adds is just another message to build and send.

- Colin


Reply via email to