[m-rev.] for review: Print version information on two lines.

[Julien Fischer]

On Thu, 6 Mar 2014, Paul Bone wrote:

> On Tue, Mar 04, 2014 at 07:23:52PM +1100, Peter Wang wrote:
>> On Tue, 4 Mar 2014 14:43:47 +1100, Paul Bone <paul at bone.id.au> wrote:
>>>> If we're going to change the output then perhaps it would be better to
>>>> just drop the word "configured"?
>>>>
>>>> Mercury Compiler, version rotd-2013-10-01, for x86_64-unknown-linux-gnu
>>>
>>> I don't mind, I don't understand the difference.  If my change is
>>> unnecessary then this change is also unnecessary.
>>
>> I didn't say otherwise.  I mooted an alternative way to satisfy your
>> goal.  It's closer to the old format and IMO it looks better, so I find
>> it easier to scan visually.  I thought it's worth consideration.
>>
>
> Okay,
>
> I think this is reasonable, and I'm not sure which 'looks better' so if you
> think this looks better I'm happy to go with that.  I also want to change
> the word "for" to "on"
>
>    Mercury Compiler, version rotd-2013-10-01, on x86_64-unknown-linux-gnu

I disagree with changing "for" to "on".  There are two possible
architectures you might report here: the host architecture (i.e.  what
mmc is running on) and the target architecture (i.e. what mmc is
generating code for).  The architecture string reported in the version
message is the latter, and "for" seems the more appropriate preposition.
For example if I have a mingw cross-compiler on Linux, mmc will (with
your change) now report:

      Mercury compiler, version FOO, on i386-pc-mingw32

even though it's not on a Windows system, but Linux.  It makes more
sense to say it's configured for i386-pc-mingw (as the message
originally did).

I suggest changing the message back to what it was -- if users are
really interested in querying aspects of the Mercury installation I
suggest implementing the options I described elsewhere in this thread.

Cheers,
Julien.