Old Boar's blog (Old Boars blog)

The interfaces project plods on. I had a meeting with the people responsible for one of the system components we're taking over/integrating on Friday. Naturally, they're a separate organisation but they are as cooperative as I could reasonably hope for. Ironically, the key components of their system are (software) modules which they inherited from the system into which we want to re-integrate them. Or, to put that another way, some function is being moved back to where it came from. This should make things simple, but of course life isn't like that. The modules which have to move have been modified, and may have dependencies in their new "host", and the original source system has been substantially redeveloped (changed database!) while they've been away. Re-integration is not going to be absolutely straightforward. This is "re-factoring" writ large. Moving function, not between objects, but between independent systems.

We are now looking at several alternative ways of bringing these "prodigal functions" back in-house. The options involve various degrees of integration, and range from bringing the whole server back, to migrating the functions to run with the "new" database.

As for timing, the main system (currently under development in India) which we have to interface with seems to have trouble understanding the idea of "non-functional requirements" in the context of batch operations. We keep on asking them when (during the batch period) they plan to send us an interface (and when they need to receive the reply). We're not talking about small files here. The stock reply is that they won't know until their system is into test. I'm sure they don't know, I'm sure I wouldn't either, but I would be aiming for something. Maybe they are just being coy.

Anyway, back to the grindstone. Not long till I escape on vacation.