Re: Weak Vectors Patch

[Clinton Ebadi]

address@hidden (Ludovic Courtès) writes:

> For these last two things, we’d need to modify the ‘vector-ref’ and
> ‘vector-set’ instructions so that they do the right thing.  The best
> solution would be to just call scm_c_vector_{ref,set_x} when
> SCM_I_WVECTP, so that the overhead remains low for regular vectors.  Can
> you look into this?

Wingo did this.

>> I'm not sure that weak vectors should have their allocated memory marked
>> as containing no pointers, but it seems to work so I left that alone in
>> case there was some magic going on that I didn't quite grok.
>
> Yes, weak vectors are to live in pointerless memory, otherwise the
> disappearing links would not disappear.
>
> The patch looks good to me, modulo a few things:
>
>   - Could you make it 3 (or 4?) different patches, each with a test case
>     showing what is being fixed?
>
>   - If that’s fine with you, you’ll need to assign copyright to the
>     FSF.  We can arrange this off-list.

I have attached a series of three patches with reformatted GNU style
commit logs and the gnu coding style violations fixed.

I talked to wingo, and he suggested that (equal? WEAK-VECTOR VECTOR) =>
#f; I have made this so by unregistering weak vectors as arrays, and
doing additional type checking in scm_eq. I am unsure if this has
ramifications for the rest of the system however.

How do you want to handle copyright assignment?

0001-Forbid-comparison-of-weak-and-normal-vectors.patch
Description: Text Data

0002-Register-disappearing-links-when-initializing-weak-v.patch
Description: Text Data

0003-Reference-weak-vectors-with-the-allocator-lock-held.patch
Description: Text Data

-- 
Corinne: this is why we should have designated bath buddies
Corinne: to get places you cant reach because youre slippery and in
         case you get a lil tooo slippery and crack your head open
         someone can call the coast guard and save you

pgpjIS3cHrJnR.pgp
Description: PGP signature