• Dan McLeran (5/5) Mar 01 2011 I am running the dmd2 compiler on my Win7 64 bit machine and everything
• Dan McLeran (3/3) Mar 01 2011 never mind, i got it. i had to pass the switches:
• Stewart Gordon (4/7) Mar 06 2011 Would you care to enlighten the rest of us on what code you were using t...
• Jonathan M Davis (14/23) Mar 06 2011 Yeah. That's weird. If you want to run the unit tests, you use -unittest...
• Stewart Gordon (17/18) Mar 07 2011
• Jonathan M Davis (11/30) Mar 07 2011 I believe that the _common_ use case for -cov is to check the coverage o...

Dan McLeran <danmcleran yahoo.com> writes:

I am running the dmd2 compiler on my Win7 64 bit machine and everything
appears to work except the -cov switch. i cannot seem to generate a .lst file.
any ideas?

thanks

dan mcleran

Dan McLeran <danmcleran yahoo.com> writes:

never mind, i got it. i had to pass the switches:

-D -unittest -cov

life is hard. it's even harder when you're dumb.

Stewart Gordon <smjg_1998 yahoo.com> writes:

On 01/03/2011 23:19, Dan McLeran wrote:
 never mind, i got it. i had to pass the switches:

 -D -unittest -cov

 life is hard. it's even harder when you're dumb.

Would you care to enlighten the rest of us on what code you were using that
requires those 
extra switches?

Stewart.

Jonathan M Davis <jmdavisProg gmx.com> writes:

On Sunday 06 March 2011 17:48:12 Stewart Gordon wrote:
 On 01/03/2011 23:19, Dan McLeran wrote:
 never mind, i got it. i had to pass the switches:
 
 -D -unittest -cov
 
 life is hard. it's even harder when you're dumb.

 
 Would you care to enlighten the rest of us on what code you were using that
 requires those extra switches?

Yeah. That's weird. If you want to run the unit tests, you use -unittest. If
you 
want code coverage, you use -unittest and -cov. If you want documentation, you 
use -D, but it has no relation to -unittest and -cov, and it isn't necessarily
a 
good idea to build your code with -D normally because of the effects that 
version(D_Ddoc) could have on your code, and there are a number places in 
druntime and Phobos which won't result in valid code if you build with -D  
(until it's fixed in the next release so that druntime and Phobos'
documentation 
is versioned with version(StdDoc) instead). So, _needing_ all three of those 
flags seems just plain wrong.

If you're using -cov, you probably want -unittest, but -D has no relation to
the 
others and shouldn't be necessary for _anything_ other than generating 
documentation.

- Jonathan M Davis

Stewart Gordon <smjg_1998 yahoo.com> writes:

On 07/03/2011 02:10, Jonathan M Davis wrote:
<snip>
 If you're using -cov, you probably want -unittest,

<snip>

I disagree:

- Most of the time I use -cov, it's to check the normal flow of program logic,
rather than 
the unittests.  It's true that it *can* be used to check that the unittests
cover all 
cases, but that's far from its only use.

- Even if checking unittests were its only *practical* use, it would be
arbitrary to 
program in a restriction explicitly.

- If there are circumstances in which code coverage analysis shouldn't be
allowed, the 
compiler should complain, not silently ignore the switch.

That it ignores -cov in the absence of these other switches is a bona fide bug.
 You're in 
a better position to put it into Bugzilla than I am.

Moreover, does DMD have a Win64 (either runs on or compiles for) version?  Or
is it just 
the normal Win32 version, running on your 64-bit system?  It would be useful if
we could 
find out the circumstances in which this bug bites.

Stewart.

Jonathan M Davis <jmdavisProg gmx.com> writes:

On Monday, March 07, 2011 06:23:09 Stewart Gordon wrote:
 On 07/03/2011 02:10, Jonathan M Davis wrote:
 <snip>
 
 If you're using -cov, you probably want -unittest,

 
 <snip>
 
 I disagree:
 
 - Most of the time I use -cov, it's to check the normal flow of program
 logic, rather than the unittests.  It's true that it *can* be used to
 check that the unittests cover all cases, but that's far from its only
 use.

 - Even if checking unittests were its only *practical* use, it would be
 arbitrary to program in a restriction explicitly.

I believe that the _common_ use case for -cov is to check the coverage of your 
unit tests, in which case you'd also need to use -unittest (that's what is 
_normally_ meant when talking about code coverage). However, there's nothing 
stopping you from using it when running your program normally.

 Moreover, does DMD have a Win64 (either runs on or compiles for) version? 
 Or is it just the normal Win32 version, running on your 64-bit system?  It
 would be useful if we could find out the circumstances in which this bug
 bites.

As far dmd itself goes, only a 32-bit binary exists, and Walter has no plans to 
create 64-bit version. In terms of compiling with dmd, Linux got 64-bit 
compilation (when you use -m64) with the latest release, but there are a number 
of things which will need to be sorted out before dmd will be able to compile 
64-bit programs on Windows (including having a 64-bit linker).

- Jonathan M Davis