[m-dev.] diff: bootstrap problem with init files in hlc grade

[Zoltan Somogyi]

On 05-Nov-2000, Fergus Henderson <fjh at cs.mu.OZ.AU> wrote:
> The cost of that in terms of code size and efficiency is trivial.
> But for the MLDS back-end I want to optimize not just the efficiency
> but also the readability of the generated code.  In terms of readability,
> the cost is small, but not entirely trivial, so it would be nice if we
> could avoid it.  Could you elaborate on what kind of future problems you're
> trying to forestall here?

Actually, one problem I am trying to forestall can be fixed *only* by
generating such a function all the time. The problem is that with the old
setup, the programmer does not usually see an initialization function being
generated, and may decide to use that name for their own initialization
(even though the name starts with mercury_, they may not know that such names
are reserved). This will work fine *until* they compile their program with
profiling.

In general, I think that not having the set of functions generated for a module
depend on the options *improves* readability.

The other kinds of problems involve the addition of other kinds of
initialization code, e.g. registration of typeclass instances.

Zoltan.
--------------------------------------------------------------------------
mercury-developers mailing list
Post messages to:       mercury-developers at cs.mu.oz.au
Administrative Queries: owner-mercury-developers at cs.mu.oz.au
Subscriptions:          mercury-developers-request at cs.mu.oz.au
--------------------------------------------------------------------------