Hi Peter
The -n flag worked fine on OS X. I don’t have this issue on Centos. As a side question I wonder why was postgres built with libedit instead of libreadline , just curious.
Back to my developer issue , he is using what he calls a data object in java. Apparently this is the place where this bad conversion happens, in other words it passes to the backend the ?? characters
Any similar trick I could use on the postgres jdbc driver ?
Will report more once I find more from him
Many thanks
Armand
> On Jan 11, 2018, at 4:12 PM, Peter Eisentraut <peter.eisentraut@xxxxxxxxxxxxxxx> wrote:
>
> On 1/11/18 16:34, armand pirvu wrote:
>> On OS X:
>> - case 1 fails
>> testdb=# insert into jt1 values ('??') ;
>> ERROR: invalid byte sequence for encoding "UTF8": 0xe2 0xa4 0x27
>> Note that at paste time ≤ changed in ??
>
> This looks like something is wrong with your libedit library. Try
> running psql with the -n option. If that helps, then look into building
> psql with libreadline instead. Because libedit is terrible.
>
>> - case 2 is fine
>> - echo -n '≤' |hexdump -C
>> 00000000 e2 89 a4 |...|
>> 00000003
>
> --
> Peter Eisentraut http://www.2ndQuadrant.com/
> PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
