Re: Where to place cmake files (Was: Help in cmake / itk issue needed)

2015-05-18 Thread Gert Wollny
Hello Andreas, 

On Mon, 2015-05-18 at 08:47 +0200, Andreas Tille wrote:
> 
> $ apt-file search /usr/share/cmake | wc -l
> 3583
> $ apt-file search /usr/lib/cmake | wc -l
> 1419

Well, when you look at the package count that provide cmake files, it
looks a bit different: 

apt-file search /usr/share/cmake | sed -e "s/:.*//"  | sort -u | wc -l 
15

apt-file search /usr/lib/cmake | sed -e "s/:.*//"  | sort -u | wc -l
48

However, the important thing, is that at least ITKConfig.cmake contains
some arch dependent lines, e.g: 

  set(ITK_FFTW_LIBDIR "/usr/lib/x86_64-linux-gnu")
 

> I was actually thinking about
> 
>git://git.debian.org/git/debian-med/bamtools.git
> 
> which I did not yet uploaded since I'm unsure about the "hand-made"
> cmake file in debian/cmake itself, the choice of a subdirectory
> debian/cmake/bamtools and the actual content of the cmake file.

A look at the cmake source code indicates that both, "lib" and "share"
are in the default search path, so I would guess, that
"/usr/share/cmake" is okay as long as there are no arch specifics in the
file (as it seems to the case with bamtools).

Best 
Gert 




-- 
To UNSUBSCRIBE, email to debian-med-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive: https://lists.debian.org/1431941328.5264.32.ca...@gmail.com



Re: Where to place cmake files (Was: Help in cmake / itk issue needed)

2015-05-18 Thread Andreas Tille
On Mon, May 18, 2015 at 11:28:48AM +0200, Gert Wollny wrote:
> > I was actually thinking about
> > 
> >git://git.debian.org/git/debian-med/bamtools.git
> > 
> > which I did not yet uploaded since I'm unsure about the "hand-made"
> > cmake file in debian/cmake itself, the choice of a subdirectory
> > debian/cmake/bamtools and the actual content of the cmake file.
> 
> A look at the cmake source code indicates that both, "lib" and "share"
> are in the default search path, so I would guess, that
> "/usr/share/cmake" is okay as long as there are no arch specifics in the
> file (as it seems to the case with bamtools).

Thanks for the hint, uploaded that way
 Andreas. 

-- 
http://fam-tille.de


-- 
To UNSUBSCRIBE, email to debian-med-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive: https://lists.debian.org/20150518100142.gb11...@an3as.eu



Re: Where to place cmake files (Was: Help in cmake / itk issue needed)

2015-05-18 Thread Ghislain Vaillant
2015-05-18 11:01 GMT+01:00 Andreas Tille :

> On Mon, May 18, 2015 at 11:28:48AM +0200, Gert Wollny wrote:
> > > I was actually thinking about
> > >
> > >git://git.debian.org/git/debian-med/bamtools.git
> > >
> > > which I did not yet uploaded since I'm unsure about the "hand-made"
> > > cmake file in debian/cmake itself, the choice of a subdirectory
> > > debian/cmake/bamtools and the actual content of the cmake file.
> >
> > A look at the cmake source code indicates that both, "lib" and "share"
> > are in the default search path, so I would guess, that
> > "/usr/share/cmake" is okay as long as there are no arch specifics in the
> > file (as it seems to the case with bamtools).
>
> Thanks for the hint, uploaded that way
>  Andreas.
>
> --
> http://fam-tille.de
>
>
> --
> To UNSUBSCRIBE, email to debian-med-requ...@lists.debian.org
> with a subject of "unsubscribe". Trouble? Contact
> listmas...@lists.debian.org
> Archive: https://lists.debian.org/20150518100142.gb11...@an3as.eu
>
>
>From my current experience with a couple of CMake-based projects, the
consensus
seems to be to put the custom CMake modules and related configs under
/usr/lib/cmake/ or /usr/share/cmake/ on Linux. I
have
seen the former path used more often than the latter.

I remembered asking about the right location of these CMake files a while
back
somewhere in d/mentors or d/science and there was no particular stands on
this.

Ghis


[MoM] r-cran-fastmatch: last polishing

2015-05-18 Thread Andreas Tille
Dear Alba,

I did some minor changes to your r-cran-fastmatch packaging:

 1. cme fix dpkg-control
As suggested in Debian Med policy this does some sensible
polishing - in this case fixing the canonical VCS fields

 2. Removed the Testsuite field from debian/control since there
is no testsuite provided

Please `gbp pull` to fetch the changes I did.

Otherwise the package is fine and thus I uploaded it.  Currently new
queue processing is quite slow and thus I can not tell when it will be
accepted in official Debian.

Thanks for your work on this

 Andreas.

-- 
http://fam-tille.de


-- 
To UNSUBSCRIBE, email to debian-med-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive: https://lists.debian.org/20150518185856.ga8...@an3as.eu



[MoM] r-cran-nnls: please fix copyright

2015-05-18 Thread Andreas Tille
Dear Alba,

I had a look into your commits for r-cran-nnls.  I noticed that there
is a file inst/COPYRIGHTS specifying copyright information for the

Files: src/*.f

Please mention this in a separate paragraph in debian/copyright since
otherwise ftpmaster will refuse the package.

Moreover I would like you to do

   cme fix dpkg-control

as I did for r-cran-fastmatch to see how helpful this command is.

Otherwise the package looks ready for upload as well.

Kind regards and thanks for your work on this

 Andreas.

-- 
http://fam-tille.de


-- 
To UNSUBSCRIBE, email to debian-med-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive: https://lists.debian.org/20150518190451.gb8...@an3as.eu



Re: [MoM] r-cran-fastmatch: last polishing

2015-05-18 Thread Alba Crespi
Dear Andreas,

2015-05-18 19:58 GMT+01:00 Andreas Tille :

> Please `gbp pull` to fetch the changes I did.
>

I've now pulled those changes.


> Otherwise the package is fine and thus I uploaded it.  Currently new
> queue processing is quite slow and thus I can not tell when it will be
> accepted in official Debian.


Thanks!

Alba


Re: [MoM] r-cran-nnls: please fix copyright

2015-05-18 Thread Alba Crespi
2015-05-18 20:04 GMT+01:00 Andreas Tille :

> I had a look into your commits for r-cran-nnls.  I noticed that there
> is a file inst/COPYRIGHTS specifying copyright information for the
>

Thanks -- I'd forgotten to add entries for that, but have done it now.


> Moreover I would like you to do
>
>cme fix dpkg-control
>
> as I did for r-cran-fastmatch to see how helpful this command is.
>

Also done.


> Otherwise the package looks ready for upload as well.
>

Let me know if you find anything else.

Alba


Re: [MoM] r-cran-nnls: please fix copyright

2015-05-18 Thread Andreas Tille
Hi Alba,

On Mon, May 18, 2015 at 08:49:15PM +0100, Alba Crespi wrote:
> > I had a look into your commits for r-cran-nnls.  I noticed that there
> > is a file inst/COPYRIGHTS specifying copyright information for the
> 
> Thanks -- I'd forgotten to add entries for that, but have done it now.

Lintian is a bit picky about this.  Did you tried

   lintian -i -I r-cran-nnls_1.4-1_amd64.changes

It gives hints to a spelling error

   wildcard-matches-nothing-in-dep5-copyright src/nnls.f  and
   unused-file-paragraph-in-dep5-copyright paragraph at line 9

A missing Copyright field

   missing-field-in-dep5-copyright

Duplicated license names dep5-copyright-license-name-not-unique
(you should rather use one paragrap "License: public-domain ..."
and refer to this).

You are no facing the most boring part of Debian packaging: handcrafting
proper copyright files. :-)

Kind regards

   Andreas.

-- 
http://fam-tille.de


-- 
To UNSUBSCRIBE, email to debian-med-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive: https://lists.debian.org/20150518202152.gc8...@an3as.eu



Re: [MoM] r-cran-nnls: please fix copyright

2015-05-18 Thread Alba Crespi
Hi Andreas,


2015-05-18 21:21 GMT+01:00 Andreas Tille :

>
> A missing Copyright field
>
>missing-field-in-dep5-copyright
>

There is no copyright holder, since it's public domain!

I can list the author there, but that's not really correct.


> Duplicated license names dep5-copyright-license-name-not-unique
> (you should rather use one paragrap "License: public-domain ..."
> and refer to this).
>

OK, I had made them separate as the "public domain" for the two files is
completely independent, not a shared licence.

Alba


Breaking ABI in shlibs (yaml-cpp)

2015-05-18 Thread Paul Novotny
I recently updated yaml-cpp from 0.5.1 to 0.5.2. This caused a bug in a
dependent package [1]. It turns out, 0.5.2 is not binary compatible with
0.5.1 (abi-compliance-checker confirms this) and is probably the source
of the bug. Does anyone have advice? What should I do to resolve this?

Thanks,
-Paul

[1] https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=784207


-- 
To UNSUBSCRIBE, email to debian-med-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive: 
https://lists.debian.org/1431994806.76158.272194361.66876...@webmail.messagingengine.com



Re: [MoM] r-cran-nnls: please fix copyright

2015-05-18 Thread Andreas Tille
Hi Alba,

On Mon, May 18, 2015 at 10:43:58PM +0100, Alba Crespi wrote:
> 2015-05-18 21:21 GMT+01:00 Andreas Tille :
> 
> > A missing Copyright field
> >
> >missing-field-in-dep5-copyright
> >
> 
> There is no copyright holder, since it's public domain!
> 
> I can list the author there, but that's not really correct.

We are now diving in the field of law where at least I feel totally
incompetent and I just repeat what I was told and what I understand with
my poor background.  There are countries (for instance Germany) where as
an author you can not "loose" your copyright as an author.  There is a
difference between copyright and a license.  To put something under
publich domain means you give permission to everybody to do whatever the
user wants to do.  However, you remain the author of the code and will
never loose this status.

>From a Debian point of view I can tell you that it is safe to list the
author under Copyright and public-domain as license.  This is what
ftpmaster wants you to do to accept the package (at least to my
experience) and I think this makes some sense to fit different law
systems.
 
> > Duplicated license names dep5-copyright-license-name-not-unique
> > (you should rather use one paragrap "License: public-domain ..."
> > and refer to this).
> >
> 
> OK, I had made them separate as the "public domain" for the two files is
> completely independent, not a shared licence.

Please push your changes to let me have a look.

Kind regards

Andreas.

-- 
http://fam-tille.de


-- 
To UNSUBSCRIBE, email to debian-med-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive: https://lists.debian.org/20150519060838.ga23...@an3as.eu



Re: Breaking ABI in shlibs (yaml-cpp)

2015-05-18 Thread Andreas Tille
Hi Paul

On Mon, May 18, 2015 at 08:20:06PM -0400, Paul Novotny wrote:
> I recently updated yaml-cpp from 0.5.1 to 0.5.2. This caused a bug in a
> dependent package [1]. It turns out, 0.5.2 is not binary compatible with
> 0.5.1 (abi-compliance-checker confirms this) and is probably the source
> of the bug. Does anyone have advice?

Quoting from the bug report: "A rebuild indeed fixes this bug" - and the
according bug #784199 is just fixed.

> What should I do to resolve this?

Keep on quoting: "... which indicates some problem with libyaml-cpp's
shlib" which might be you could do something about - but I admit I
personally don't know what.  Besides librime1 there is another rdepends
librivet11 which should be probably checked whether it is affected as
well.  You might consider asking either at
debian-ment...@lists.debian.org or yaml-cpp upstream (or both) whether
there might be some explanation and fix for this issue.

Hope this helps

  Andreas.

-- 
http://fam-tille.de


-- 
To UNSUBSCRIBE, email to debian-med-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive: https://lists.debian.org/20150519062220.gb23...@an3as.eu