Skip to content

Examples to unittest [part 5]#4049

Closed
wilzbach wants to merge 1 commit intodlang:masterfrom
wilzbach:examples_to_unittest5
Closed

Examples to unittest [part 5]#4049
wilzbach wants to merge 1 commit intodlang:masterfrom
wilzbach:examples_to_unittest5

Conversation

@wilzbach
Copy link
Contributor

@wilzbach wilzbach commented Mar 4, 2016

Last part of my sweep throught std for replaceable docstring unittests.

There is still one pattern that might be incorporated in the future - multiple examples with text between them, e.g.


Some intro ....
.... Amazing module ....

assert(a == 1)

.... another use case ....

assert(a == 1)

@wilzbach wilzbach force-pushed the examples_to_unittest5 branch from cca1b16 to 535f388 Compare March 4, 2016 02:40
@wilzbach
Copy link
Contributor Author

wilzbach commented Mar 4, 2016

Apparently the change in uni.d causes auto-tester to fail - seems a bit weird ...
Has anyone an idea?

@DmitryOlshansky
Copy link
Member

Apparently the change in uni.d causes auto-tester to fail - seems a bit weird ...
Has anyone an idea?

I recall a lot of dark magic involved to work around forward reference errors and link failures in std.uni, this might be triggering some of unresolved bugs in the compiler still.

@wilzbach
Copy link
Contributor Author

splitted into std.socket to #4111 to isolate the failing problem.

@wilzbach
Copy link
Contributor Author

can someone of the admins tag this with "FP magic from dmd"?

@9il
Copy link
Member

9il commented Mar 25, 2016

How this PR is related to Floating Point (FP)?

@9il
Copy link
Member

9il commented Mar 25, 2016

Looks like I need to change the label name

@wilzbach
Copy link
Contributor Author

How this PR is related to Floating Point (FP)?

Oh I thought FP means "Function pointer" (or sth. similar). Sorry!
It still fails due to some magic related to dmd, though ...

@9il
Copy link
Member

9il commented Mar 25, 2016

It still fails due to some magic related to dmd, though ...
Hide all checks

I don't think so. This is because we have bad build scripts. I had the same problem with another PR. Probably our build scripts link old library when use new source code. So linker can not find a new name.

@wilzbach
Copy link
Contributor Author

I don't think so. This is because we have bad build scripts.

So how do we move forward with this problem? On my machine make -f posix.mak std/uni.test passes.

@9il
Copy link
Member

9il commented Mar 29, 2016

So how do we move forward with this problem? On my machine make -f posix.mak std/uni.test passes.
Hide all checks

No hope for now

@quickfur
Copy link
Member

Maybe try rebasing onto latest master and see if that makes a difference? It's a long shot, but maybe it's an unrelated problem or something broken in Phobos that got fixed since?

@quickfur
Copy link
Member

FWIW, I ran the entire Phobos test suite on my machine (Linux/64bit), and there were some compile errors (not the same as the autotester failure though), but after rebasing onto master, the testsuite passed with flying colors. Just sayin'.

@wilzbach wilzbach force-pushed the examples_to_unittest5 branch from f56c984 to 3907ffc Compare March 30, 2016 04:54
@wilzbach
Copy link
Contributor Author

Maybe try rebasing onto latest master and see if that makes a difference?

Done. Let's see & hope for the best :)

@wilzbach
Copy link
Contributor Author

Okay the problem is that d_do_test.d (part of dmd, run by make -f posix.mak auto-tester-test) uses std.uni and the linking doesn't work.

I opened an issue at dmd and will close the issue now until this is fixed in dmd.

@wilzbach wilzbach closed this Mar 30, 2016
@9il
Copy link
Member

9il commented Mar 30, 2016

I opened an issue at dmd and will close the issue now until this is fixed in dmd.

Thanks!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

5 participants