I have created this INFRA ticket regarding .net-2.0, node labels and the
missing .net-3.5 on windows-2012-3:

https://issues.apache.org/jira/browse/INFRA-14645

2017-07-19 10:54 GMT+02:00 Dominik Psenner <[email protected]>:

>
>
> On 19 Jul 2017 6:35 a.m., "Stefan Bodewig" <[email protected]> wrote:
>
> On 2017-07-18, Dominik Psenner wrote:
>
> > oohh how I like to be the bringer of good news! Right now I was finally
> > able to fix the building of the netstandard target by providing a
> > jenkinsfile script that detects the UID and GID of the working path and
> > passing that in as parameters into the builder-netstandard dockerfile
> when
> > it is built. That dockerfile then creates a user with that UID and GID
> such
> > that the nant executable finds a place to store its cached assemblies and
> > what else it needs.
>
> Wow, congrats. Many thanks for your persistence.
>
> > The following todos are yet to complete before building and testing the
> > assemblies is ok:
>
> > * node 'windows-2012-3' [1] lacks the installation of .net framework 3.5
> > whereas 'windows-2012-1' [2] and 'windows-2012-2' [3] have it (probably
> > needs to be fixed by INFRA)
> > * build and test assemblies for net-2.0 (probably needs to be fixed by
> > INFRA by installing that on the windows nodes)
>
> It may happen that it's not possible to install 2.0 once a more recent
> framework is there, I'm not sure.
>
>
> If thats the case then net-2.0 may be installed on windows-2012-3. Gotta
> verify that. In any case it would be sensible if infra would add additional
> labels on jenkins nodes so that the labels reflect the nodes' capabilities.
> Something like dotnet-4.5, dotnet-4.0, dotnet-3.5, ... What do you think?
>
>
> > * verify and probably fix testing of the assemblies for target
> net-standard
> > * parse the test results when testing net-standard
> > * the old build job can probably safely be disabled/removed because the
> new
> > pipeline does the same builds and regression tests and even more (please
> > discuss)
>
> +1 for removing the old build.
>
>
> We still have to find and fix what's causing the old test job to fail,
> though.
>
>
> Stefan
>
>
>


-- 
Dominik Psenner

Reply via email to