[Arsperl-users] Problems writing to join tables - no entry ID to return

[Tim S. <tsk...@st...>]

        A short summary: I'm having problems with ARS::ars_CreateEntry ()
returning a useful return value when it is creating an entry in a join
form/table.  This appears to be a systemic problem, and I'm not really
sure what to do about it.  I'm curious if anybody else has found a
solution to this.
        
        Some background: I'm our current resident ARSperl developer at
Stanford.  We use Remedy to handle our Incident Management and other
related areas, and we're starting to build a CMDB based on the same
infrastructure.  I've built up the general interface to handle all of
this, based on (at the lowest level) the ARSperl interface.

        I'm currently working on the engine to populate the BMC CMDB
product based on spec CMDBf data (this is XML, for those that don't know).
The CMDB is based very heavily on join tables; pretty much *everything*
you write to is a join table of some sort or another, and there's no
real way to work around this.  

        However, our testing has found a major problem with this: it is
impossible through the API to write directly to a join table.  Instead,
when information is offered to the join table (via ars_CreateEntry ()),
underlying work-flow creates the appropriate entries *indirectly*.  This
generally works - that is, the entries are created appropriately, both in our test
forms and in the default BMC_CMDB forms.  But the correct value is not
returned, according to the API docs:

        http://www.arsperl.org/manual/ars_CreateEntry.html

        On success, we're supposed to get a request ID.  On failure, we
get undef.  In this case, we get a *blank* request ID.  All we can really
say about this is that we can tell the difference between a failure and a
success (though in this case, it leaves us in a situation of "we know we
wrote, but we don't know *where* we wrote", which I would argue subverts
the API pretty badly).

        So - what to do about it?  I've started writing up some wrapper
code in my own libraries that, if we get a blank return code, goes out and
tries to find the entry again based on the data offered during the create
(dying if 0 or 2+ entries are found); but this doesn't seem like the right
general solution.  This seems like something that the ARSperl libraries
should take care of on its own, though I have to admit that I don't really
know how it *would* do so.

        Advice, thoughts, and code examples are welcome.  I'm more than
happy to elaborate as well.

                            - Tim Skirvin (tsk...@st...)
-- 
   Information Technology Services      http://www.stanford.edu/~tskirvin/
System Software Developer, Unix Team           Stanford University