version(all) block in src/phobos/unittest.d unnecessary?
There's a version(all) block in src/phobos/unittest.d (not std.unittest) with this comment: // Bring in unit test for module by referencing function in it and then a bunch of random function calls from each module. I've tried commenting out the code in the version block and call make unittest, and from what I can tell the unittests will still run. Is this block unnecessary?
Re: version(all) block in src/phobos/unittest.d unnecessary?
On 28.05.2011 19:01, Andrej Mitrovic wrote: There's a version(all) block in src/phobos/unittest.d (not std.unittest) with this comment: // Bring in unit test for module by referencing function in it and then a bunch of random function calls from each module. I've tried commenting out the code in the version block and call make unittest, and from what I can tell the unittests will still run. Is this block unnecessary? unittest.d seems like an artifact from older times. I think win32 makefile needs a rewrite to behave more like posix.max (posix one tests both release and debug builds and all modules one by one). -- Dmitry Olshansky
Re: version(all) block in src/phobos/unittest.d unnecessary?
What the heck, I've just opened the linux makefile and it's beautiful compared to the windows one. The windows one is 3x the size. Is this because Linux uses GNU's make perhaps?
Re: version(all) block in src/phobos/unittest.d unnecessary?
On 28.05.2011 19:54, Andrej Mitrovic wrote: What the heck, I've just opened the linux makefile and it's beautiful compared to the windows one. The windows one is 3x the size. Is this because Linux uses GNU's make perhaps? Don't think so, though admittedly I haven't looked very deep. One reason is that win32 had a collection of workarounds for optlink. And that not counting the sheer verbosity of it. Me thinks that if anyone has the skill to bring to shape win32 makefile that would make an awesome pull request. -- Dmitry Olshansky
Re: version(all) block in src/phobos/unittest.d unnecessary?
On 5/28/2011 9:03 AM, Dmitry Olshansky wrote: On 28.05.2011 19:54, Andrej Mitrovic wrote: What the heck, I've just opened the linux makefile and it's beautiful compared to the windows one. The windows one is 3x the size. Is this because Linux uses GNU's make perhaps? Don't think so, though admittedly I haven't looked very deep. One reason is that win32 had a collection of workarounds for optlink. And that not counting the sheer verbosity of it. Me thinks that if anyone has the skill to bring to shape win32 makefile that would make an awesome pull request. It is due to the relative primitiveness of the make.exe that walter ships with the windows version of things. It's.. weak.
