[m-rev.] for review: pack small args next to local secondary tags

[Julien Fischer]

On Tue, 10 Jul 2018, Julien Fischer wrote:

> On Tue, 10 Jul 2018, Zoltan Somogyi wrote:
>
>>  On Mon, 9 Jul 2018 11:33:25 +1000 (AEST), Julien Fischer
>>  <jfischer at opturion.com> wrote:
>>>  There's one issue with them: the classes in C# and Java corresponding to
>>>  du_ptag_layouts require either (1) the addition of an extra argument to
>>>  their constructors or (2) the code generators to be updated to not emit
>>>  the extra argument (since it won't do anything in those grades anwyay).
>>>
>>>  Mercury/css/array.cs(194,55): error CS1729: The type
>>>  `mercury.runtime.DuPtagLayout' does not contain a constructor that takes
>>>  `4' arguments
>>>  ../runtime/mercury_dotnet.cs(738,12): (Location of the symbol related to
>>>  previous error)
>>>  Mercury/css/array2d.cs(52,45): error CS1729: The type
>>>  `mercury.runtime.DuPtagLayout' does not contain a constructor that takes
>>>  `4' arguments
>>>  ../runtime/mercury_dotnet.cs(738,12): (Location of the symbol related to
>>>  previous error)
>>
>>  The attached diff *should* fix this problem. Julien, could you please
>>  review it (it is short), and try it out?
>
> I've committed a small fix to the C# version.  The libraries now compile
> and install in both the C# and Java grades on my system.  (I'll do full
> bootchecks of both grades this evening.)

Building the stage 3 library dependencies is causing the compiler to
loop in both the C# and Java grades, however this is not anything to do
with this change.  (The same thing is happening for the
hard_coded/lexer_ints test case.)

I'll go back through the rotds and work out what change introduced
this behaviour.

Julien.