lib/cache: type safety of the cache API pointers

added refactoring label

changed milestone to %5.2.0

mentioned in merge request !1025 (merged)

writing this properly without libknot would need 10-20 additional lines of code

I had written that before choosing the shorter committed variant, so I might just dump it here for reference:

	struct lmdb_env *env = db2env(db);
	/* Gather required information.  For more confidence we don't use
	 * our "cached" env->mapsize and we use a fresh transaction. */
	int ret = cdb_commit(db, stats);
	MDB_envinfo info;
	if (!ret)
		ret = lmdb_error(mdb_env_info(env->env, &info));
	MDB_txn *txn = NULL;
	if (!ret)
		ret = txn_get(env, &txn, true);
	MDB_stat st;
	if (!ret)
		ret = lmdb_error(mdb_stat(txn, env->dbi, &st));

	if (!ret) {
		free_txn_ro(env); // the TXN might be broken right now
		return kr_error(ret);
	}

	/* Formula copied from knot_db_lmdb_get_usage() but it's really simple. */
	const size_t db_size = st.ms_psize
		* (st.ms_branch_pages + st.ms_leaf_pages + st.ms_overflow_pages);
	return (double)(100 * db_size) / info.me_mapsize;

approved this merge request

@vcunat Please rebase this MR and I will review it again. It has conflicts with !1042 (merged).

assigned to @vcunat

unapproved this merge request

added 48 commits

• ae3acc00...0c8d0bd4 - 46 commits from branch master
• d1d8cc2f - lib/cache: type safety of the cache API pointers
• 3e375558 - lib/cache kr_cdb_api::space_usage(): also use kr_cdb_pt

Compare with previous version

assigned to @pspacek and unassigned @vcunat

We ended up doing much more cache changes than I anticipated when creating this MR. I think it's good now.

added 6 commits

• 3e375558...3a6c03b4 - 4 commits from branch master
• 9bcdeff9 - lib/cache: type safety of the cache API pointers
• 9b469602 - lib/cache kr_cdb_api::space_usage(): also use kr_cdb_pt

Compare with previous version

approved this merge request

enabled an automatic merge when the pipeline for 9b469602 succeeds

mentioned in commit f742d3bc

merged

mentioned in merge request !1060 (merged)

mentioned in issue #611 (closed)