Le 11/04/2022 à 20:37, Nathan Bossart a écrit :
> On Fri, Apr 08, 2022 at 05:16:06PM +0200, Gilles Darold wrote:
>> Attached v7 of the patch that should pass cfbot.
> Thanks for the new patch! Unfortunately, it looks like some recent changes
> have broken it again.
>
>> +enum trivalue schema_is_exclude = TRI_DEFAULT;
>> +
>> +/*
>> + * The kind of object filter to use. '0': none, 'n': schema, 't': table
>> + * these values correspond to the -n | -N and -t command line options.
>> + */
>> +char objfilter = '0';
> I think these should be combined into a single enum for simplicity and
> readability (e.g., OBJFILTER_NONE, OBJFILTER_INCLUDE_SCHEMA,
> OBJFILTER_EXCLUDE_SCHEMA, OBJFILTER_TABLE).
>
>> * Instead, let the server decide whether a given relation can be
>> * processed in which case the user will know about it.
>> */
>> - if (!tables_listed)
>> + if (!objects_listed || objfilter == 'n')
>> {
>> appendPQExpBufferStr(&catalog_query, " WHERE c.relkind OPERATOR(pg_catalog.=) ANY (array["
>> CppAsString2(RELKIND_RELATION) ", "
> I think this deserveѕ a comment.
>
Attached v8 of the patch that tries to address the remarks above, fixes
patch apply failure to master and replace calls to pg_log_error+exit
with pg_fatal.
.Thanks.
--
Gilles Darold