Re: [RFC/RFA] Per-objfile data mechanism

[David Carlton <carlton at kealia dot com>]

On Sun, 10 Aug 2003 21:03:02 +0200 (CEST), Mark Kettenis <kettenis@chello.nl> said:
> From: David Carlton <carlton@kealia.com>
> Date: Tue, 15 Jul 2003 09:48:31 -0700

>> I was also going to write, based on a cursory misreading of Mark's
>> patch, that it simplified memory management in some circumstances,
>> but now that I look at it more closely, I think I just misread the
>> patch.  (I may still be misreading the patch; my head is spinning
>> with other things.)  Would it be possible/beneficial to modify the
>> mechanism to provide an optional per-datum cleanup function as
>> well?

> I quite deliberately left per-datum initializers and destructors out
> to encourage the use of the per-objfile obstacks.  But they can
> always be added if they're needed.

The concrete reason for that suggestion is that I have a patch
awaiting review adding some per-objfile data that consists of an
expanding hash table; that can't be handled with an obstack.  In
general, I get the feeling that we're moving a bit more to data
structures that are less obstack friendly, but who knows.  Having said
that:

> So what's the final verdict.  Should my patch go in, or do people
> have concrete ideas about necessary improvements or alternative
> implementations?

I certainly wouldn't want to stand in the way of putting it in now: if
we do decide we want to add per-datum cleanup mechanisms to your
patch, we can do it just as easily after the patch has been applied as
before the patch has been applied.

David Carlton
carlton@kealia.com