On 13/12/2025 22:36, Coty Sutherland wrote:
On Sat, Dec 13, 2025, 2:23 PM Mark Thomas <[email protected]> wrote:

12 Dec 2025 16:15:07 Christopher Schultz <[email protected]>:

In your opinion does tcnative have any decent kind of unit-test
coverage?

No.

We don't necessarily need to have Java-based unit tests, but even
C-based tests?

No.

I only ask because it would be very handy to be able to perform a quick
smoke-test of a tcnative release on a few platforms without having to
stand-up Tomcat, etc. I know your setup is quite elaborate and it's
fairly "easy" for you to do all of this, but my environment is much
more modest. I'd still like to cast meaningful votes for tcnative, even
if they are compile-and-unit-test votes only.

All I have is various combinations of OpenSSL and Tomcat Native that I
then run the Tomcat unif tests with.

I'm not sure if C based tests would be worth the effort.


I don't think it would be (but I'm not great with C either), but what about
java unit tests and integration tests that run after the library is built
to load and test it? I was playing around with this some after Chris's
reply earlier and found it to be viable (I think). I also found that
calling one of the functions reliably causes a jvm crash when called with
the test, but doesn't seem to be used by tomcat (can't recall which
function it was ATM). Will look into that again Monday if I can find time.

Isn't that what the Tomcat unit tests do?

If you are able to fund out how to trigger that JVM crash that would certainly be interesting.

I do need to update Gump to run better combinations.

Or / and? maybe we ask for a Jenkins instance and build all the
interesting combinations and then test them with Tomcat.


That's probably a good idea either way. I had one setup in my homelab but
the machine died and I never stood it back up :/

Something else to add to the ever growing TODO list.

Mark


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to