Alexander Lamaison <swish_at_lammy.co.uk> writes:
> On 10 March 2010 09:55, Simon Josefsson <simon_at_josefsson.org> wrote:
>>
>> Btw, maybe we can add an libssh2_session_error_last with size_t in the
>> next version, and deprecate libssh2_session_last_error? Then at least
>> one interface will be fixed...
>
> Why bother?
>
> Fixing one call isn't going to solve the overall problems so why make
> extra work for clients when we could just wait till the soname bump?
> Also session_last_error is the perfect name for this function;
> session_error_last sounds awful.
I was just thinking if we can start this process without a soname bump,
by replacing each and every API call with a new one that has size_t in
it. Then we could officially deprecate the old interfaces, and after
some time make the old APIs call syslog and warn about obsolete
interface, and after some more time, do the soname bump and remove the
old interfaces.
But maybe biting the dust and fixing this problem for real, and do a
soname bump, is better.
We could also rename the header file, to make it possible to have both
API/ABI versions installed at the same time.
I don't feel strongly about what to do, but I feel something should be
done so that we aren't stuck with a sub-optimal design.
/Simon
_______________________________________________
libssh2-devel http://cool.haxx.se/cgi-bin/mailman/listinfo/libssh2-devel
Received on 2010-03-10