Re: RFS: viennacl
Hi Anton
On 04/08/2011 07:43 AM, Anton Gladky wrote:
> Hi, Michael,
>
> I am not a DD, but maybe my 2 cents will help you to make your package better:
>
> 1. Why is it experimental-oriented? I have built your package with
> sid, and it builds normally.
Yes, it will build normally, because it doesn't compile anything, it
just repacks the tar-ball with some additional files, essentially ;-).
But as you can see, the libviennacl-dev package Depends: on libopencl1
and opencl-headers which are only in experimental. I don't think it
would make much sense having a source package in unstable whose binary
packages depend on experimental, or am I wrong?
> 2. It would be good if you add a "check" section to your package, if
> it is possible. Please, see eigen2 [1] for that. They start
> verification tests, but even if they fail, the packaging succeed. Also
> before tests they output cpu and memory information [2]. I do not
> know, will your tests will work on servers without GPU. But it is a
> good practice to add verification tests.
Yes, I was thinking about that. That would require me to also
Build-Depends: on above mentioned packages, though. I will give it a try
today and see how it works out.
>
> Very interesting library! Probably, will test it for my tasks.
I thought so too ;-) I'm not sure though, how much efficiency they lose
by dispatching kernels for every operation individually, though, and the
benchmarks they publish are not very informative IMHO.
>
> Anton
>
> [1] http://git.debian.org/?p=pkg-kde/kde-req/eigen2.git
> [2] http://git.debian.org/?p=pkg-kde/kde-req/eigen2.git;a=blob;f=debian/rules;h=c3f1c492ce152332077b3f14f82d08b61ae211ba;hb=a66f03be1f080344f374637bc32bb76b485b3472#l14
>
Thanks for your time and input
Michael
>
> On Fri, Apr 8, 2011 at 6:43 AM, Michael Wild
> <themiwi@users.sourceforge.net> wrote:
>> Dear Debian Science Developers,
>>
>> As already mentioned on the list, I'm packaging ViennaCL, a scientific
>> C++ header-only library based on OpenCL, and now I am looking for a
>> sponsor for my package "viennacl".
>>
>> * Package name : viennacl
>> Version : 1.1.1-1
>> Upstream Author : Karl Rupp <rupp@iue.tuwien.ac.net>
>> * URL : http://viennacl.sourceforge.net
>> * License : MIT/X11 (Expat)
>> Section : contrib/science
>>
>> It builds these binary packages:
>> libviennacl-dev - Scientific computing library written in C++ based on
>> OpenCL
>> libviennacl-doc - ViennaCL API and user documentation
>>
>> The package appears to be lintian clean, only in pedantic mode it
>> complains about a single utility binary that has been left in the source
>> tar-ball by upstream.
>>
>> The upload would fix these bugs: 620135
>>
>> My motivation for maintaining this package is that I am using it in my
>> research and wanted to get some practice packaging with a relatively
>> simple project before I submit a much more involved package.
>>
>> The package can be found on mentors.debian.net:
>> - URL: http://mentors.debian.net/debian/pool/contrib/v/viennacl
>> - Source repository: deb-src http://mentors.debian.net/debian unstable
>> main contrib non-free
>> - dget
>> http://mentors.debian.net/debian/pool/contrib/v/viennacl/viennacl_1.1.1-1.dsc
>>
>> The Git repository is on alioth.debian.net:
>> - Browser: http://git.debian.org/?p=debian-science/packages/viennacl.git
>> - Clone: git+ssh://git.debian.org/git/debian-science/packages/viennacl.git
>>
>> I would be glad if someone reviewed and uploaded this package for me.
>>
>> Kind regards
>>
>> Michael Wild
>>
>>
>> --
>> To UNSUBSCRIBE, email to debian-science-request@lists.debian.org
>> with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
>> Archive: [🔎] 4D9E9289.8060204@users.sourceforge.net">http://lists.debian.org/[🔎] 4D9E9289.8060204@users.sourceforge.net
>>
>>
>
>
Reply to: