Currently the shell processes --eval prior to any .js files on the command line. Is there any particular reason this order is used? I find it confusing and backwards to what I expect, probably because I think of --eval as a non-interactive replacement for the interactive REPL (unless --shell is used), and when .js files are specified without --eval they are loaded prior to presenting the prompt.

My use case is a library of code contained in a .js file which defines some functions that I can then call from --eval. The code isn't general enough that I want to have it always loaded in my ~/.mongorc.js file. I can work around it by using load() from inside --eval, but that's ugly.

$ cat foo.js
print("foo");
foo = function () { print("bar"); }
$ mongo foo.js --eval 'foo(); print("baz")'
MongoDB shell version: 2.4.9
connecting to: test
Wed May 28 12:16:43.912 ReferenceError: foo is not defined
$ mongo --eval 'load("foo.js"); foo(); print("baz")'
MongoDB shell version: 2.4.9
connecting to: test
foo
bar
baz
$