This is the mail archive of the guile@sources.redhat.com mailing list for the Guile project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: Again: Guile + Boehm's gc


Dirk Herrmann <dirk@ida.ing.tu-bs.de> writes:

> So, I hope that it brings something new to you.

Thanks a lot for writing this all down!

I think most of your points are valid, but I would weight them
differently I guess.  I don't see it as problematic that you have to
help the GC here and there by telling it about roots, for example.

Indeed, I think that most of the good things you see in the Boehm
collector might go away when it is tuned for Guile.  I expect the the
tuning will include making the collector more precise and less
conservative.
 
> * No separation of SCM cell and corresponding malloced area.  For
>   example with strings, vectors, bignums etc. guile uses the following
>   approach:

The extra indirection is helpful when you want to grow or shrink a
vector without changing its identity, for example.  I don't know if we
are doing it right now, but I think this is an important feature to
have.  With a conservative GC, you can not move objects around like
you can we a precise one.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]