Brendan Jurd wrote:
> On Tue, Jan 27, 2009 at 3:25 PM, Brendan Jurd wrote:
> > On Tue, Jan 27, 2009 at 3:05 PM, Tom Lane wrote:
> >>
> >> strncpy is generally deprecated; any remaining uses you find of it
> >> are probably only there for lack of round tuits. Use strlcpy in new
> >> code, unless
On Tue, Jan 27, 2009 at 3:25 PM, Brendan Jurd wrote:
> On Tue, Jan 27, 2009 at 3:05 PM, Tom Lane wrote:
>>
>> strncpy is generally deprecated; any remaining uses you find of it
>> are probably only there for lack of round tuits. Use strlcpy in new
>> code, unless there's a pretty strong argument
On Tue, Jan 27, 2009 at 3:05 PM, Tom Lane wrote:
> Brendan Jurd writes:
>> A quick grep through the backend code shows that strlcpy and strncpy
>> are both in use, with neither having a clear majority. I used strncpy
>> because it is more prevalent within src/backend/utils/adt.
>
> strncpy is ge
On Mon, Jan 26, 2009 at 8:32 AM, Alex Hunsaker wrote:
>
> -the various (only moved not added)
> strncpy(...)
> copy[len] = '\0';
>
> just use strlcpy?
>
Thanks for having a look Alex.
strlcpy would be easier, but I thought there might be portability
concerns re its non-entirely-standard status.
On Mon, Jan 19, 2009 at 15:26, Brendan Jurd wrote:
> From the better-late-than-never department, comes a patch to improve
> the handling of AM/PM and AD/BC markers in to_timestamp(), and up the
> ante on error reporting for various kinds of invalid input.
Ok cool I tried this out and gave it a lo
