Conversation
This will let us more cleanly enable/disable individual sub-tests in the Phantom branch.
This will allow us to turn tests on/off without affecting each-other. In general making individual tests independent of each-other makes them easier to understand and debug.
robgjansen
left a comment
There was a problem hiding this comment.
My main concern is that we'll be rewriting some tests twice, once for glibtest and again for rust. But I suppose that's OK if we don't want to pull in Rust yet at this point.
Yeah, tweaking the tests to use the glib framework is probably faster than rewriting them in Rust; at least faster than adding the first Rust code/test, but I agree overall it's adding an extra step if we eventually want to migrate to Rust. Maybe with this test merged so that you have something to work with, I'll take a look at using Rust for the udp test. |
I guess this introduces a larger question about plans for our next release. Do we intend to release phantom changes before we start migrating the wider Shadow code-base to Rust? Or should we keep the phantom changes in a [public] dev branch as we continue improving Shadow in other ways, but without making a formal release? If we do the former, we may not want to introduce Rust as yet another dependency yet. But if we do the latter, which I'm in favor of, I would say we could go ahead with Rust whenever we feel ready. |
|
Suppose we create a |
|
Seems reasonable, but let's discuss at our meeting next week before taking action. That way we can include Ryan in the discussion. |
Enables individual file tests to be turned on and off independently, so that we can enable them incrementally in the phantom branch.
Migrates to glib's test framework so that we can filter which tests run from the command-line. This also makes the test code more succinct and give better failure reporting.
Modifies subtests to be independent of each-other; i.e. not depend on earlier tests to have created or modified a test file.