[Proj] ntf_r93.gsb file corrupted

[Cuinet Jérôme]

Franck, thanks for these answers.

Fortunalately there are regression tests, quickly regarding source code I 
have not seen them !

The "mandatory" term it isn't for maintenance. Updates concerns improvements 
or corrections, no damaging ! Damages are on bêta corrections, on trunk 
development tree. Release is an implicit declaration that the evolvement are 
validated. Releases are not a "gamma version", and my system administrator 
says me that update to the newest version is mandatory. I'm heretic if I 
don't update my system.

It is principles that I think correct.

I know that no-regression detections are an hard task. But we can't damage a 
software, we can also do improvements. Some users trust open source 
developpers. If they can't, if an open source software is an always "gamma 
software", a task is really never closed, we must say it !

I realize that this mail seems to be a troll, but I only want how I can 
trust in open source software, and especially in proj4 library.

If that I think is wrong, so I will know that and I will not be able to 
believe a software description :(


Best regards

Jérôme Cuinet


----- Original Message ----- 
From: "Frank Warmerdam" <warmerdam at pobox.com>
To: "PROJ.4 and general Projections Discussions" <proj at lists.maptools.org>
Sent: Thursday, September 24, 2009 8:49 PM
Subject: Re: [Proj] ntf_r93.gsb file corrupted


Cuinet Jérôme wrote:
> A question :
>
> Are not there some coordinate transformation tests which verifiy the
> calculations ?
>
> I don't know how often mistakes appears, but this type of tests don't 
> seems
> to be diificult to construct.

Jérôme,

There are regression tests, but they do not include ntf_r93.gsb even now.
I would add the problem corrupted the file in a relatively subtle way that
would only be detected for a grid shift in some places.

In fact I see there is a testIGNF but it was commented out in the past when
it wasn't obvious to determine whether we had the supporting gsb file or 
not.

I am updating the Makefile.am to include the testIGNF and testntv2 checks
if the corresponding grid shift files seem to be available.  I'll produce a
new beta later today with these changes.

Note these are only run on unix systems.

 > A no-regression test could have been mandatory, no ? Some unusual 
functions
 > might be broken without anybody pick out it.

The word "manditory" has limited application in open source.  It is very 
rare
that I get paid for any PROJ.4 maintenance, and I'm only doing this because
when I took it on there was no one else willing to do it.

There is certainly room for additional PROJ.4 developers and contributors.

Best regards,
-- 
---------------------------------------+--------------------------------------
I set the clouds in motion - turn up   | Frank Warmerdam, 
warmerdam at pobox.com
light and sound - activate the windows | http://pobox.com/~warmerdam
and watch the world go round - Rush    | Geospatial Programmer for Rent

_______________________________________________
Proj mailing list
Proj at lists.maptools.org
http://lists.maptools.org/mailman/listinfo/proj