[ADMB Users] Tests and recommendations for ADMB

Ian Taylor - NOAA Federal ian.taylor at noaa.gov
Tue Nov 26 18:04:00 PST 2013

Hi Jon,
Thanks for your thorough write-up of your installation efforts and the list
of recommendations. I think that your review document touches on a number
of key issues facing the ADMB project.

I obviously can't speak for the project as a whole, but can provide some
information on a few of your points.

First of all, I agree that everything is more complicated than it should
be. I think there's hope that this process will be simpler. However, keep
in mind that Johnoel was asking for feedback about something that's in
development rather than a polished official release.

In theory, ADMB has a "release
which would have a core team of developers thoroughly test all releases and
update the documentation before each official released. This would protect
the average user, septuagenarian or otherwise, from any broken code or
incomplete documentation. However, it's my impression that the ADMB Core
Team <http://www.admb-project.org/developers/core-team> is too small and
too busy with other stuff to do all the work that needs to be done under
this procedure. Also, I get the feeling that not everybody involved with
ADMB does well with the idea of procedures in general. So the result has
been a large number of discussions about ADMB development occurring on the
users list including requests for help with testing things that are less

Regarding changes between versions, there are files CHANGES.txt and
NEWS.txt that are included in the distribution (I'm don't remember if
there's a reason to have both). CHANGES.txt is also linked in the most
recent new version
the ADMB website. Many of these changes were discussed at length on the
ADMB Developer's email list (archived
and others on the ADMB Users list (archived
In particular, having "safe" be the default compile method has been
suggested for years.

I doubt that the day will come when a single installation procedure will
work for everybody. I think the best approach will be describing in detail
the simplest installation method possible for any operating system, which
would be recommended as the default, while also providing space on the ADMB
website for documentation of alternative procedures. This could be
accompanied by a table discussing the pros and cons of each approach.
Definitely the out of date installation instructions on the website could
use some cleaning up.

Thanks again for the feedback--hopefully your recommendations will lead to
some improvements.

On Sun, Nov 24, 2013 at 12:38 AM, Jon Schnute <schnutej-dfo at shaw.ca> wrote:

> Johnoel and Dave (and many others) – At last I’ve managed to build ADMB in
> Windows with mingw. Great work making this possible! I’m genuinely
> impressed.
> After this initial step, I decided to test things more completely. I
> enclose a report that describes my results and suggestions in complete
> detail. We still have bugs in the scripts adcomp and adlink, and I really
> hope these can be fixed. Also, changes from past versions of ADMB make the
> current version confusing. For example, the script admb.cmd now has
> different arguments than it once did.
> So I’ve used this occasion to pull my experiences together in hopes that
> we can make ADMB easier to install, understand, and use. I hope you, other
> developers, and general users find them useful.
> I have one more suggestion not mentioned in the enclosed file. Please also
> explain how to obtain a 64-bit version of mingw and build a 64-bit version
> of ADMB.
> Best wishes to all,
> Jon
> PS – I’m willing to help with the documentation after we get everything
> working and I understand more completely how the package is supposed to
> work.
> _______________________________________________
> Users mailing list
> Users at admb-project.org
> http://lists.admb-project.org/mailman/listinfo/users
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.admb-project.org/pipermail/users/attachments/20131126/392b30ea/attachment.html>

More information about the Users mailing list