libmysqlclient soname version bumped by upstream

[Michał Piotrowski]

2011/3/21 Tom Lane <tgl at redhat.com>:
> =?ISO-8859-2?Q?Micha=B3_Piotrowski?= <mkkp4x4 at gmail.com> writes:
>> 2011/3/21 Tom Lane <tgl at redhat.com>:
>>> Now I have no problem with pushing such a change into rawhide, but what
>>> shall I do about Fedora 15?  It doesn't seem very nice to force a soname
>>> bump after alpha.  On the other hand, the alternatives seem worse:
>>>
>>> * never upgrade F15 past mysql 5.5.9
>
>> Not very interesting solution :)
>
> I agree, particularly since 5.5.x is still pretty immature --- there
> are likely to be significant bug fixes coming down the pike.

Backporting security/important fixes can be very painful.

>
>>> * revert F15 to 5.1.x release series
>
>> Please _don't_ do this.
>
> No, I don't want to do that either.  Also, on reflection it would
> *still* require a recompile to do that, since we hacked some of the
> exported symbols in the process of reducing libmysqlclient's list of
> exposed symbols.  So that choice seems fairly useless.
>
> We might be able to get away with forcing the old soname version to
> continue being used, but that would mean F15 diverges from pretty
> much everybody else.

Can wrong ABI version cause problems in other apps?

>
> On the whole I think a forced recompile of dependent packages might be
> the best solution, painful as it is.
>
>                        regards, tom lane
> --
> devel mailing list
> devel at lists.fedoraproject.org
> https://admin.fedoraproject.org/mailman/listinfo/devel
>



-- 
Best regards,
Michal

http://eventhorizon.pl/