Open stylewarning opened 11 years ago
This is a tentative proposal for a solution.
It would be better to have the utility computation functions always loaded (that is a constant amount of bloat, which should be okay for the end user), and then have a generated utility database file, which can be read in. (This should not be extremely difficult since only strings and symbols are used.) That way, we don't need this overly baroque way of loading/unloading ASDF, but rather just loading and unloading a small database file.
The only troubles are designing such a file, writing the serializer/parser for it, and making sure it gets distributed.
Right now, the database of utilities, as well as the utility computation functions, are loaded and unloaded on demand. This is potentially memory leaky (though actual benchmarks don't indicate any pragmatic concerns), very costly, and very wasteful. It would be better to do it in another way.
I'm marking this as an enhancement primarily, but also as a bug because loading/unloading ASDF is potentially memory leaky.
This issue addresses some concerns in Issue #19.