Re: mdnblocks() sabotages error checking in _mdfd_getseg()

Robert Haas <robertmhaas@gmail.com> writes:
> The comment in mdnblocks.c says this:

>                          * Because we pass O_CREAT, we will create the
> next segment (with
>                          * zero length) immediately, if the last
> segment is of length
>                          * RELSEG_SIZE.  While perhaps not strictly
> necessary, this keeps
>                          * the logic simple.

> I don't really see how this "keeps the logic simple".

My (vague) recollection is that this is defending some assumptions
elsewhere in md.c.  You sure you haven't broken anything with this?
Relation extension across a segment boundary, perhaps?
        regards, tom lane