Fedora Rawhide (27)

First one:
It seems, you stopped exproting the "my_init()" symbol. Is that expected?
Is it true that its functionality is now provided by mysql_init ?

Solving this one here rhbz#1470642, take a look at the attachement.

Another question is about load_defaults symbol – my understanding is that people used load_defaults basically for loading additional groups in most cases, which is correctly done by mysql_options(conn, MYSQL_READ_DEFAULT_GROUP, "<mygroup>"); now – is that explanation correct or at least close to correct?

The load_defaults / my_load_defaults thing actually seems rather odd to me: those symbols aren't documented as having been removed in 10.2, AFAICS, and they're in the same header file in 10.1 and 10.2 (private/my_default.h) . Yet somehow they're available when building against 10.1 and not when building against 10.2.

(also, my_getopt.h - which is a public header - includes my_default.h - which is a private header...I filed MDEV-13327 on that).

Yes, the patch (from the rhbz#1470642 comment 10) looks correct.

mschorm, shall I keep this issue open for more Fedora libmariadb.so.3 issues? Or it's ok to close it?

At this point, Fedora bug tracker for libmariadb.so.3 issues has been fully resolved and closed.
I still have an open tracker for Fedora change - using library from C/C instead of server one (using separate packages) and I have very few issues open there, which could be potentialy related.

I'd suggest either to:
drop this JIRA issue priority to Minor and leave it open
or
close this issue and if I'd hit some, I'll open a new issue.

I'm equally fine with both. The issues seems to be fixed, but I still have work for weeks ahead, which could be pontentialy related.