max_allowed_packet variable limits the size of a single result set. In the
[mysqld] section, any normal connection can only get
1M worth of data in a single query.
mysqldump you typically produce "extended INSERT" queries, where you list multiple rows within the same
INSERT command. It's better, then, to have this variable set high.
The above is a brief explanation. Now, as for your question "What is the point?" -- there is no clear answer. The
my-XXXXX.cnf set of files are long since outdated, and have their issues, including recommendations which are commonly not agreed upon (like the amount of memory you should allocate to the
innodb_buffer_pool_size). So take it as a general starting point. In particular, these settings relate to systems with very low memory available (like
128M RAM), so the numbers don't really apply to today's hardware.
One could argue that you won't necessarily import the dump into same server, so possible to another server with higher setting for
As a side note, I usually set
max_allowed_packet to very high values, like
64M. As opposed to other variables like
max_allowed_packet you only pay for what you need. So a connection will not allocate the full
64M if it doesn't have to.
1M limit means you could have some row with some
BLOB, possibly larger than
1M, and in such case, you would actually be unable to read that row.