Re: Updating ortools

2024-01-23 Thread Drew Parsons

On 2024-01-23 14:02, Kari Pahula wrote:

Hi, I requested access to
https://salsa.debian.org/science-team/ortools some time ago.  Is
anyone available to process it?

I have a prospective (experimental) changes available at
https://salsa.debian.org/kaol/ortools.  It requires versions of abseil
and protobuf from experimental, at least the CI build was apparently
unable to pull dependencies from it.


done



Re: Updating ortools

2024-01-23 Thread Kari Pahula
Hi, I requested access to
https://salsa.debian.org/science-team/ortools some time ago.  Is
anyone available to process it?

I have a prospective (experimental) changes available at
https://salsa.debian.org/kaol/ortools.  It requires versions of abseil
and protobuf from experimental, at least the CI build was apparently
unable to pull dependencies from it.



Re: Updating ortools

2024-01-11 Thread Kari Pahula
On Wed, Jan 10, 2024 at 12:55:22PM +0200, Kari Pahula wrote:
> I've been looking at the new version of ortools 9.8 but looks like
> retaining Python support with it would take significant effort.  The

Even with skipping Python support for now, the Abseil version in
unstable is currently too old.  But there's a transition coming soon
(https://bugs.debian.org/1059535) to it so I think the best is just to
wait until it hits sid.

I'll continue with this then.



Re: Updating ortools

2024-01-10 Thread Kari Pahula
On Tue, Jan 02, 2024 at 08:48:49PM +0100, Anton Gladky wrote:
> sure, welcome to team! it would be also good to fix RC bugs also there. 
> Thanks!

I've been looking at the new version of ortools 9.8 but looks like
retaining Python support with it would take significant effort.  The
new ortools uses https://github.com/pybind/pybind11_protobuf whose
cmake files are designed around just downloading dependencies and
among them it uses an include file called proto_api.h from pybind11,
whose use is deprecated by the upstream and not installed anymore and
according to https://bugs.debian.org/1060225, that's by design and the
current way of doing it is by using something called ubp,
https://github.com/protocolbuffers/protobuf/tree/main/upb, which
pybind11_protobuf doesn't use yet.

This is about as far as I'm ready to follow up with this thing.  I
would ask to just drop Python support for now and return to this once
upb is available in Debian (I don't quite know how that'd happen since
upstream explicitly say that they won't make any releases of that
software) and pybind11_protobuf doesn't use any deprecated not
installed files it only gets because it downloads and doesn't even
consider using releases or system installed files.

If there's a consensus that Python support is a must then I guess we
could stick with an old version of ortools.  But I'd rather update
without Python for now and return it later when this stuff matures a
bit more.  The current state is that it FTBFS universally and isn't in
testing or installable in current unstable either so it's not like
people could have used it as it is.



Re: Updating ortools

2024-01-02 Thread Anton Gladky
Hi,

sure, welcome to team! it would be also good to fix RC bugs also there. Thanks!

Regards,

Anton

Am Di., 2. Jan. 2024 um 20:15 Uhr schrieb Kari Pahula :
>
> Hi.
>
> As the maintainer of minizinc, I have an interest in ortools.  I added
> ortools-flatzinc as rdep for it with a recent update but noticed that
> it's kind of unmaintained at the moment.
>
> I can prepare an update with the newest upstream version and add
> myself as an uploader.
>



Updating ortools

2024-01-02 Thread Kari Pahula
Hi.

As the maintainer of minizinc, I have an interest in ortools.  I added
ortools-flatzinc as rdep for it with a recent update but noticed that
it's kind of unmaintained at the moment.

I can prepare an update with the newest upstream version and add
myself as an uploader.