We debated this a while back and those of us who wanted a formal
specification got flamed for wanting to be "spoon-fed" an implementation in
our language-of-choice.
My organization has encountered a number of pitfalls along the way, and I'd
be glad to impart some of my wisdom on to you, answer your questions, or
help you out however I can. Be forewarned though, we don't use a lot of
OpenSRS's commands, just the ones we need. Makes sense, huh?
A good example of a foible that should be documented: When registering a
domain, "address2" is entirely optional but if you DO send it, and it's
something like "#30" it will be rejected because "at least one alpha
character is required" which seems a bit odd on an optional field, we had to
do a good bit of manual data correction to fix those... Lots of little
quirks in their system and I too wish I would have had the benefit of a more
detailed spec.
The answer you are going to get is that "There is a spec and it's written in
Perl" - one side will tell you that Perl is a nice generic language that's
easy to pick up on - which is very true. My argument was that if my
implementation of the bleachboy-protocol on which the internet runs was
written in PDP-11 machine code, that would hardly constitute a "spec".
Regardless, I'd be glad to help you out, just mail me specific questions. :)
peace
.---------------------------------------------------------------.
| bleachboy bboy at bboy dot net +1 (615) 260-4931 |
| ICQ 1839892 UNIX: Because you want to USE your computer |
`---------------------------------------------------------------'
> -----Original Message-----
> From: owner-dev-list@opensrs.org [mailto:owner-dev-list@opensrs.org]On
> Behalf Of jim@archer.net
> Sent: Wednesday, March 22, 2000 9:49 AM
> To: Brad Murray; James H. Cloos Jr.
> Cc: jim@archer.net; dev-list@opensrs.org
> Subject: I think we need a more detailed spec
>
>
> Thanks Brad, that's very helpfull. I missed the part about the 16bit
> int. Perl is not my strong suit.
>
> I'm wondering if anyone has prepared an API spec for the Open SRS
> server? I read, several times, the one thats on the site, but I think
> we need something a little more specific. We should be able to code
> to the spec, in any language, without having to refer to perl scripts
> or any other source code.
>
> Ideally, I think the spec should:
>
> 1 - describe the structure of a generic command. So, if all commands
> must be preceeded by 16 bits of data describing the length of what is
> to follow, then it should say that and describe which endian flavor
> is expected. (For example, In Perl, what exactly is a "packed int"?)
>
> 2 - describe the structure of a generic reply in the same manner.
>
> 3 - include a complete list of commands, with each commands syntax,
> and the possible replies for each command.
>
> 4 - describe exactly what encryption algorathim is used and exactly
> which parts of each command are encrypted.
>
> That said, I am willing to help write such a spec (I have a lot of
> experience doing that). Prehaps we could all work on it on this list?
> Maybe someone could post what they have as a starting point (if
> anyone has anything)? I will volunteer to cycle peoples comments into
> it, clean it up, post it, and repeat until we all agree we have
> something that can be understoon and is complete.
>
> In the meantime, I would appreciate it if anyone could post a
> description of how to talk to the server or else post code in C/C++
> or Java that can do it (assuming there is any).
>
> Thanks!
>
> Jim
>
>
>
>
> --- Original Message Follows ---
>
> The protocol is essentially binary. If you'll look more carefully at
> the perl scripts, you'll see that a packed 16bit integer must come
> before any command, data, etc. This integer tells the client/server
> how
> much data to expect before processing the command. The protocol
> itself
> is very simple, but the issue is getting encryption working. I know
> that bboy@bboy.com has build a library in C that properly communicates
> with the server over an encrypted channel (I think he's using
> libdes).
> I've tried doing the same thing using PHP and libmcrypt with no
> success. I did have a completely functional client going without
> encryption, but don't know if it would work with the current system.
>
> Brad
>
>
> >
> > |> I'
This archive was generated by hypermail 2.1.3 : Tue Oct 19 2004 - 23:35:22 EDT