We need circuit launch and cannibalization unit tests

[dgoulet]

In particular, these functions circuit_launch_by_extend_info() and circuit_find_to_cannibalize() really needs unit tests!

They do so much and they are so core that we need to assess what they are doing is ok. See #24469 for an example of what it should have done.

[nickm]

Per our triage process, these tickets are pending removal from 0.3.4.

[nickm]

These tickets, tagged with 034-removed-*, are no longer in-scope for 0.3.4. We can reconsider any of them, if time permits.

[ffmancera]

I have been working on a test. I think I came with a good approach but the test is giving me an error. I don't know if the function tested is just not working properly or the test is wrong.

Please check my github ​branch bug25118! Thank you =)

[dgoulet]

Ok so basically here is the problem:

May 04 09:31:55.761 [debug] circuit_launch_by_extend_info(): Haven't fetched enough directory info yet; canceling circuit launch.

A useful trick for you is to use ./src/test/test <your-test> --debug which will output the debug logs of the test. You can see that you don't get at all in the function but rather exit very fast with a NULL.

So couple of things:

1. Use tt_assert(new_circ) to make sure you do have a circuit that you expected.

2. router_have_minimum_dir_info() is mockable so you'll have to mock it to return true.

This circuit_launch_by_extend_info() is really not simple to unit test, you might need to mock lots of things. So instead, you could skip it and just unit tests circuit_find_to_cannibalize() which it self will be something intense as well.