Request to use GNU guix reproducibility bugs data for research project | University of Waterloo

Request to use GNU guix reproducibility bugs data for research project | University of Waterloo

[zimoun]

Hi,


On Fri, 29 Oct 2021 at 16:48, Muhammad Hassan <muhammad.hassan@uwaterloo.ca> wrote:

> I would like to scrap reproducibility bugs data from the linked
> website to use in my research project that is being conducted at the
> University of Waterloo. I am a Master's student.

Nice!  Thank for your interest to Guix.


> The project aims to provide automated support for non-reproducibility
> detection.

Non-reproducibility can be large. Against which one are you working?


> Please tell me if there is an API that I can use to scrap the data.

For instance, “guix build --no-substitutes” followed by “guix build
--check” allows you to locally build and rebuild for checking if the
build is reproducible – bitwise speaking.  Then, “guix challenge” allows
to check your local builds against the build-farm.  Note that two
build-farms are available [1,2] for various architectures.

The Guix project runs a Data Service to collect various data. This
service serves a JSON API (not-really documented but roaming on IRC
#guix helps ;-)).  One year ago (already, damned!), a tiny and drafty
script to somehow get the reproducibility coverage was around [3].  I
have not worked on it since then and I do not know if it still works.
Perhaps, it might be an entry-point for working on reproducibility data
from Guix.

Feel free to ask more details.


1: <http://ci.guix.gnu.org/>
2: <http://bordeaux.guix.gnu.org/>
3: <https://lists.gnu.org/archive/html/guix-devel/2020-10/msg00181.html>


Hope that helps,
simon

PS: I close the issue since it is not one, IIUC.  Instead, let redirect
to the mailing list guix-devel which seems a better place for discussing
such request. :-)

Re: Request to use GNU guix reproducibility bugs data for research project | University of Waterloo

[Julien Lepiller]

[-- Attachment #1: Type: text/plain, Size: 2323 bytes --]

The guix data service also has some info on reproducibility. See data.guix.gnu.org
(Select master, the latest processed revision, and add /package-reproducibility to the URL). This page compares between berlin and bordeaux.

There are a lot of unknowns because the build farms haven't both built some of the packages, so we can't compare, but of those tgat have been built on both sites, we can already observe some interesting differences.

Le 2 novembre 2021 08:01:51 GMT-04:00, zimoun <zimon.toutoune@gmail.com> a écrit :
>Hi,
>
>
>On Fri, 29 Oct 2021 at 16:48, Muhammad Hassan <muhammad.hassan@uwaterloo.ca> wrote:
>
>> I would like to scrap reproducibility bugs data from the linked
>> website to use in my research project that is being conducted at the
>> University of Waterloo. I am a Master's student.
>
>Nice!  Thank for your interest to Guix.
>
>
>> The project aims to provide automated support for non-reproducibility
>> detection.
>
>Non-reproducibility can be large. Against which one are you working?
>
>
>> Please tell me if there is an API that I can use to scrap the data.
>
>For instance, “guix build --no-substitutes” followed by “guix build
>--check” allows you to locally build and rebuild for checking if the
>build is reproducible – bitwise speaking.  Then, “guix challenge” allows
>to check your local builds against the build-farm.  Note that two
>build-farms are available [1,2] for various architectures.
>
>The Guix project runs a Data Service to collect various data. This
>service serves a JSON API (not-really documented but roaming on IRC
>#guix helps ;-)).  One year ago (already, damned!), a tiny and drafty
>script to somehow get the reproducibility coverage was around [3].  I
>have not worked on it since then and I do not know if it still works.
>Perhaps, it might be an entry-point for working on reproducibility data
>from Guix.
>
>Feel free to ask more details.
>
>
>1: <http://ci.guix.gnu.org/>
>2: <http://bordeaux.guix.gnu.org/>
>3: <https://lists.gnu.org/archive/html/guix-devel/2020-10/msg00181.html>
>
>
>Hope that helps,
>simon
>
>PS: I close the issue since it is not one, IIUC.  Instead, let redirect
>to the mailing list guix-devel which seems a better place for discussing
>such request. :-)
>
>

[-- Attachment #2: Type: text/html, Size: 3207 bytes --]

Re: Request to use GNU guix reproducibility bugs data for research project | University of Waterloo

[Vagrant Cascadian]

[-- Attachment #1: Type: text/plain, Size: 801 bytes --]

On 2021-11-02, Julien Lepiller wrote:
> The guix data service also has some info on reproducibility. See
> data.guix.gnu.org (Select master, the latest processed revision, and
> add /package-reproducibility to the URL). This page compares between
> berlin and bordeaux.

Thanks, just added a link at https://reproducible-builds.org/citests/

Been meaning to do that for ages...


> There are a lot of unknowns because the build farms haven't both built
> some of the packages, so we can't compare, but of those tgat have been
> built on both sites, we can already observe some interesting
> differences.

Yeah, the number of untestable packages looks a little low...

As of now, on x86_64:

    Matching (7210, 34.76%)
    Not matching (1748, 8.43%)
    Unknown (11784, 56.81%)


live well,
  vagrant

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 227 bytes --]