[enh] Issue 8765 - Assert should print expression when no message is present #1203
Conversation
|
I like more informative assert output, but I don't think that the .di files should be impacted. Your changes are too early in the lifecycle and instead it should be moved to very late, nearer code generation. |
|
I think we should also print the failing expression when the error message is present. |
|
@braddr: Agreed. int x;
assert(x == 1);
assert(x == 1, "x is not 1")core.exception.AssertError@test.d(6): assert(x == 1)
core.exception.AssertError@test.d(7): assert(x == 1), "x is not 1"I think pulling this will avoid having to write assert messages in most cases since they will be much more informative by default. |
|
I have no idea why these two fail: http://d.puremagic.com/test-results/pull.ghtml?runid=338676
The tests worked on my win32 machine. |
|
I'm gonna have to get a vbox machine for a x64 linux system, I don't know why some platforms are failing (mostly x64 + OSX 32bit). |
|
I can't build Phobos on x64 posix, the build either runs indefinitely or runs out of memory (in vbox). If someone has the time and skill set to figure out what's wrong I'd kindly ask if they could have a look (test report: http://d.puremagic.com/test-results/pull.ghtml?runid=340630), thanks. |
| @@ -2004,6 +2004,26 @@ elem *AssertExp::toElem(IRState *irs) | |||
| Type *t1 = e1->type->toBasetype(); | |||
|
|
|||
| //printf("AssertExp::toElem() %s\n", toChars()); | |||
| OutBuffer buf; | |||
| Expression* msgExpr; | |||
There was a problem hiding this comment.
And this declaration should be moved down to where it's initialized.
|
It's possible your bug is using |
|
Maybe, but it only fails on x64 systems. I'll try your suggestion though. Edit: Actually wrong it also failed on x32. We'll see what the autotester says soon. |
|
It seems to almost consistently fail with x64 (-m64) binaries (except OSX which fails all the time). @WalterBright: Sorry to bother you, do you have any idea what could be causing this? |
|
And now it's working again, how strange. Let's see if it works again -- restarting tester. |
…re function `hasMutableIndirectionParams()` is based on the old `TypeFunction::purityLevel()` (before fixing issue 8408). So its result is consistent with `TypeFunction::purityLevel() != PUREstrong` in 2.060. Then it *fixes* the regression.
Issue 9177 - Wrong `alias func this` incorrectly reports error message
Also updated test cases. By a recent change in 2.061 (4b2767e), direct returning of NewExp sometimes avoids immutable implicit conversion. So, test5081 should return mutable array values through local variables.
Issue 9230 - Incorrect implicit immutable conversion occurs in pure function
Issue 9234 - DMD confuses string template parameter with function
Issue 9237 - Add isPOD trait
ghost
closed this
specified. Fixes Issue 9255 - Assert needs use file name for the AssertError .file field.
ghost
reopened this
ghost
closed this
|
Uhhhh what is up with Github? I see one thing in |
|
I'll try to open a new pull request. |
ghost
mentioned this pull request
This makes this code:
Print:
core.exception.AssertError@test.d(2): assert(1 == 2)But I can't figure out a way to write a test-case because the test is compilable but it's expected to fail at runtime. I need to compare the runtime message just like diag-style tests in the fail_compilation folder. Any help there is appreciated.