Re: [Gcl-devel] GNU make bug

[Vadim V. Zhytnikov]

Camm Maguire writes:
> Greetings!
>
> "Vadim V. Zhytnikov" <address@hidden> writes:
>
>
> > Camm Maguire writes:
> >
> > > Greetings, and thanks Vadim!  This is now committed.
> > > You asked earlier about a 2.7.0 roadmap.  I'll try to put together a
> > > comprehensive list soon, but obviously the top is ansi compliance.
> > > I'm wondering if I can solicit you help in proposing some patches to
> > > close some of our ansi test suite failures.  If so, do you have an
> > > area with which you feel most comfortable?  I'd really appreciate
> > > someone concentrating on the formatter/printer if possible.  The idea
> > > of course is to preserve the spirit of the code in place.
> > > Take care,
> >
> > Hi Camm!
> >
> > I've been testing recent gcl 2.7.0 twc with Maxima
> > and observe several Maxima test failures and some
> > performance degradation.  First of all I'd like
> > to consier these issues.  Formatter/printer is OK
>
>
> Thank you for looking into this!  I'd suggestion using the tag
> Version_2_7_0t1.
>
> There are two possible performance related issues of which I am
> aware.  
>
> 1) We have a smaller default hole size which expands with the core
>    proportionally.  This should add a bit of time purely in gc, and
>    can be circumvented by setting the hole size.
>
> 2) There was an array compilation problem that has since been cleared
>    in the above mentioned tagged version.

OK, I'll look into this shortly (I hope :-)

> > for me but I can't really tell when I could
> > dive into the problem - recently my schedule
> > for this summer hardened quite a bit.
> > But I really don't know it may be really
> > not so bad ...
>
>
> Would be very helpful!
>
> I'm hoping we can release by summer's end, but I don't think we should
> until the overwhelming majority of Paul's ansi tests pass.

Probably I can do something substantial only in August.


--
     Vadim V. Zhytnikov

      <address@hidden>
     <address@hidden>