Conversation
|
Fails on CI because runner can't download DMD: |
|
Nice, you've added a place for GC unit tests (they used to get put at the bottom of src/rt/lifetime.d.) |
Wow good catch! |
|
Rebased and now the tests are passing. Except the code coverage test, though. However, this just adds tests, and I don't know how it could this affect the rwmonitor.d and thread.d coverage, given that it should at least increase the coverage (not that it's touching any of that). |
|
Oh, that actually looks like it's showing the lines that are not covered, not delta. I guess it just has more lines with these tests, so the ratio is smaller. |
|
Not a bad idea to start a separate test suite for the GC. There's a problem with the test folder, though: it's never built on Windows. Similar unittests are currently in src/gc/impl/conservative/gc.d. If you actually care about this test only, it might be better to just add it there. |
|
Ah, wasn't aware of that. I'll fix it tonight, thanks!
…On 10 May 2017 9:01 am, "Rainer Schuetze" ***@***.***> wrote:
Not a bad idea to start a separate test suite for the GC. There's a
problem with the test folder, though: it's never build on Windows.
Similar unittests are currently in src/gc/impl/conservative/gc.d. If you
actually care about this test only, it might be better to just add it there.
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
<#1679 (comment)>, or mute
the thread
<https://github.com/notifications/unsubscribe-auth/AA9weVmsA9bZ1F6lWcqP8fllpls4LQ0fks5r4WDcgaJpZM4Kc4Aa>
.
|
|
|
@rainers I've updated the code, so that I've moved the tests just into the unittest in the suggested file. Thanks! |
Issue 9275 describes the runtime crashing on the invalid pointer passed to removeRoot, instead of documented no-op. This has been fixed in the meantime, so unittest is added as a guarantee that it stays so.
Calling
removeRooton something that was not addedby addRoot should be possible.
https://issues.dlang.org/show_bug.cgi?id=9275