On Fri, Sep 07, 2018 at 11:22:14PM +0200, Fabien COELHO wrote: > Weird indeed. However, even if the patch applied cleanly for me, it was not > compiling anymore. Attached an updated version, in which I also tried to > improve the error message on trailing garbage.
At least now it applies for me, thanks. + Integer values expected for keywords <literal>port</literal>, + <literal>connect_timeout</literal>, + <literal>keepalives_idle</literal>, + <literal>keepalives_interval</literal> and + <literal>keepalives_timeout</literal> are parsed strictly. Wouldn't it be better to actually document what "parsed strictly" means? A wild though: we already have things like pg_strtoint32 or pg_atoi which do similar error handling in smarter ways. Wouldn't we want to refactor the code so as libpq makes use of such routines? This would mean that the error string should be moved around, and allowing frontends to use those utilities requires some extra engineering. Not that this patch should reinvent the world... -- Michael
signature.asc
Description: PGP signature
