> A while ago it was being held that the Postgres large object data type
> was too new and not sufficiently tested and mature to be used in a
> production environment. I am about to deploy a little database that
> involves storing large-ish text files (20-500k) which could be either done
> as large objects or as pointers to those files that would be then stored
> as ordinary files in the OS's filesystem. I am undecided as to this
> question. What are the pros and cons? What is the performance in either
> case vis-a-vis the other? It is a web app; the interface is done in PHP.
> Beginning from which version is the large object interface (if at all)
> to be considered stable and ready for production?
>
I would always use a reference solution. It's database independent
and if the contents you store is URL, then you may easily get the
referenced text file.
Marten