Re:

[Ferdinando M. Ametrano-2]

At 03:05 PM 12/11/01 +0000, you wrote:
>[ Sorry, resending, headers were mangled --edd ]
They're still mangled. I don't think quantlib-dev will receive the message.
I append Dirk's message below (you may want to read his message before my
reply ;-)

>new versions of -quantlib and -ruby
>are going into the archive in 7 hours.
No -python? Why rebuilding -quantlib?

>Shall we do it one arch at a time?
do you mean -quantlib and -ruby before, -python later?

>   I am still open to even preventing builds
>on some of the less useful archs (arm, m68k, mips, mipsel) but maybe that
>should be the very last resort.
If 0.1.9 compiled on those platform I would try with 0.2.1

>Don't we have a few weeks left before 0.3.0?
I _hope_ 0.3.0 will get out late January. Given past experiences this could
mean late February

ciao -- Nando

=======================

>[hidden email], Wouter Verhelst" <[hidden email]>
>MIME-Version: 1.0
>Content-Type: text/plain; charset=us-ascii
>Content-Transfer-Encoding: 7bit
>From: Dirk Eddelbuettel <[hidden email]>
>To: Ferdinando Ametrano <[hidden email]>
>Cc: "[hidden email]"@eddelbuettel.com,
>         Wouter Verhelst" <[hidden email]>
>Subject: Re: [Quantlib-dev] Re: The build...
>In-Reply-To: <[hidden email]>
>References: <[hidden email]>
>         <[hidden email]>
>         <[hidden email]>
>X-Mailer: VM 6.92 under 21.4 (patch 1) "Copyleft" XEmacs Lucid
>
>
>[ Sorry, resending, headers were mangled --edd ]
>
>   "Nando" == Ferdinando Ametrano <[hidden email]> writes:
>   Nando> sorry for the late reply, but I am catching up with 200 unread
>   Nando> messages .....
>   Nando>
>   Nando> Two changes: 1) I've just added to the QuantLib-Python/Ruby SWIG
>   Nando> interface files a check that should ensure the correct version of
>   Nando> QuantLib is used:
>   Nando>
>   Nando> #if QL_HEX_VERSION < 0x000300a3 #error using an old version of
>   Nando> QuantLib, please update #endif
>
>Perfect -- I was going to suggest exactly that.
>
>   Nando> I could create a new CVS snapshot at
>   Nando> http://quantlib.org/snapshot.html tomorrow, tagged as 0.3.0a3
>   Nando>
>   Nando> 2) Dirk, what about removing the SWIG tool dependency from
>   Nando> QuantLib-Python and QuantLib-Ruby Debian packages? The wrap files
>   Nando> generated by the QuantLib developers with their own specific choice
>   Nando> of SWIG version should be used instead of re-generating them.
>
>Actually, I have done that last night; new versions of -quantlib and -ruby
>are going into the archive in 7 hours.
>
>   Nando> Dirk, could we use the 0.3.0a3 dev-only release with the two changes
>   Nando> above for the bug-hunting?  Which platform should we start with?
>   Nando> Alpha?  Can you provide us an account somewhere?  We could compile
>
>Yes, I should have news about an Alpha account for Luigi shortly.
>
>It might be that the header mismatch fixes most if not all the bugs. Let's
>wait a day or two to let the build daemons react to 0.2.1-2 for -ruby and
>-quantlib.
>
>   Nando> QuantLib 0.3.0a3 on those platforms, but who should we ask to
>   Nando> install it in order to compile QuantLib-Python/Ruby?  Sorry if I'm
>   Nando> missing some obvious Unix-Debian issue because of my ignorance
>
>Shall we do it one arch at a time?  I am still open to even preventing builds
>on some of the less useful archs (arm, m68k, mips, mipsel) but maybe that
>should be the very last resort.
>
>Don't we have a few weeks left before 0.3.0?
>
>Dirk
>
>
>--
>Good judgment comes from experience; experience comes from bad judgment.
>                                                             -- F. Brooks

[Dirk Eddelbuettel]

On Tue, Dec 11, 2001 at 03:59:59PM +0100, Ferdinando Ametrano wrote:
> At 03:05 PM 12/11/01 +0000, you wrote:
> >[ Sorry, resending, headers were mangled --edd ]
> They're still mangled. I don't think quantlib-dev will receive the message.

Oh boy. My success rate at these between-bed-and-morning-shower emails is
terrible. I should rethink this approach.

> I append Dirk's message below (you may want to read his message before my
> reply ;-)
>
> >new versions of -quantlib and -ruby
> >are going into the archive in 7 hours.
> No -python? Why rebuilding -quantlib?

Sorry, typo:  -python and -ruby, of course.

> >Shall we do it one arch at a time?
> do you mean -quantlib and -ruby before, -python later?

No, sorry, build architecture. As it will typically involve getting an
account and doing some digging, it's best to do it one at a time.

Let's start with Alpha, and let's see if the other fall in line.  I'm all
for concentrating on meaningful archs (alpha, sparc, ia64, s390, ...)

> >  I am still open to even preventing builds
> >on some of the less useful archs (arm, m68k, mips, mipsel) but maybe that
> >should be the very last resort.
> If 0.1.9 compiled on those platform I would try with 0.2.1

Fair point.

> >Don't we have a few weeks left before 0.3.0?
> I _hope_ 0.3.0 will get out late January. Given past experiences this could
> mean late February

Ok. I am sure we can sort the portability issues out before then. Would be
nice to cover seven, eight, nine, ... of the twelve available architectures
[ some have currently somewhat-broken toolchains (hppa), some have fewer
available machines (arm, m68k) so hitting all of them is not very likely ]

Dirk

--
Good judgment comes from experience; experience comes from bad judgment.
                                                            -- F. Brooks