Re: [Sqlrelay-discussion] SQLRelay 0.37pre5 is not backward compatibile

[David M. <dav...@fi...>]

Yeah, there's just the one line:
	uses (u)int[16|32]_t instead of short/long, etc.

Which covers some changes I made to the C/C++ api's.  Those changes
percolated up to the other api's too.  But, you're right, it's not very
explicit.  And there isn't a line about changing rowCount to a 64 bit
integer from a 32 bit.  I'll add some lines like that.

In C/C++, on 32-bit machines, a "long" is 32 bits and a "long long" is
64 bits.  On 64-bit machines, both are 64 bits.  I was running into
problems in a  64bit-host/32bit-client (or vice-versa) environment, so I
had to switch "long" to int32_t and "long long" to int64_t.  While I was
at it, I switched "short" to int16_t, got rid of "int" references (which
could be either 16 or 32 bits depending on the platform) and made a few
values 64 bit, such as the row count.

Sorry about the incompatibility, but I don't imagine it will change
again before 1.0

Dave

On Tue, 2006-01-31 at 12:58 +0100, Maciej Wisniowski wrote:
> Hi
> 
> I've now switched to SQLRelay 0.37pre5 and
> found some changes that caused my old code was not
> working.
> These are changes in SQLRConnection constructor (port
> is now short not int), and in rowCount (returns long not int).
> I think these things should be listed in changelog, or maybe
> they are there but in a inplicit way?
>