DBLib clarity

[PlainName]

Oh, well thanks for pointing me in the right direction 

[plazma]

I built a component database library to my workplace. The unique identifier is a company internal code received from the purchase department. Each manufacturer part number got its own internal code. In practice I could have used the manufacturer part number as the unique identifier.

I'm building my hobby database library on exel. There is not so many codes so excel sheet works fine. For this database library I use the manufacturer part number as the unique identifier.

Each component can have multiple footprints. Most have three: IPC Least, Nominal and Most. Some have a fourth footprint which is usually manufacturer specified.

So for example a part with QFN and SOIC footprints appears twice in the library. That's because the QFN and SOIC parts have their own manufacturer part numbers. And both of them can have multiple footprints. Like QFN can have IPC L, N and M. Also the SOIC got the different IPC densities.

[PlainName]

I could have used the manufacturer part number as the unique identifier

Seems to something of a favorite, but I wonder how appropriate it is. A major issue is that there is no guarantee that manufacturers will have unique numbers between them, but another is that the format varies wildly.

Instead, why not use a guaranteed unique to you number? The formatting will be consistent and appropriate (if you wanted, you could ensure no 0 or 1, so you don't confuse them with O and l), so you know they will fit in the space provided, and you won't end up with 40-character-wide columns showing 6-character numbers.

[Pseudobyte]

Most people use an incrementing index for the unique identifier...I think?

Would anyone be interested in seeing how to setup an AWS RDS (postgresql) deployment for Altium DBlib? If that is valuable even to a few people I can write up a post.