On Fri, Feb 18, 2011 at 2:47 PM, Peter Geoghegan
<peter.geoghegan86@gmail.com> wrote:
> On 17 February 2011 19:28, Dave Page <dpage@pgadmin.org> wrote:
>> No objections to that, but it does need to handle interoperability
>> with older versions too. IOW, it needs to read/write booleans as
>> REG_SZ values of "true" or "false", not REG_DWORDS. There may be other
>> gotchas too - that's the one I was hitting though.
>
> What about the Read() overloads? They currently work by passing a
> variable by reference (in the C sense, by pointer), to an overload of
> read. ISTM that it would be preferable to have variants with different
> names that return variables by value. What do you think?
I'm not sure that helps - the issue is that the current code reads the
string "true" and treats it as a boolean. What we need to do is ensure
we always write booleans as strings, otherwise it won't be backwards
compatible. Then, ReadBool() just reads a string value, and converts
it.
> By the way, this isn't that much trouble. I've already finished with
> the write() overloads. We're mostly using the inline wrapper functions
> in sysSettings.
OK.
--
Dave Page
Blog: http://pgsnake.blogspot.com
Twitter: @pgsnake
EnterpriseDB UK: http://www.enterprisedb.com
The Enterprise PostgreSQL Company