[OpenSPIM] Best practices for using a different camera?

[Johannes Schindelin]

Hi Ulrik,

On 2015-01-22 14:57, Ulrik Günther wrote:

>> On Jan 22, 2015, at 2:51 PM, Johannes Schindelin 
>> <johannes.schindelin at gmx.de> wrote:
>> 
>> We could (for lurkers, Ulrik speaks about a server dedicated to 
>> building 
>> and testing the software, say, every night). There is a Jenkins CI at 
>> http://jenkins.imagej.net/ that Fiji and ImageJ use. We could easily 
>> set 
>> something up there.
>> >
>> 
>> But we do not have any regression tests; therefore, the usefulness of 
>> such a continuous integration is not yet large enough to merit the 
>> huge 
>> amount of work to get the continuous integration up and running *and* 
>> to 
>> keep it running.
>> 
> 
> Well, until we have actual tests we could at least see whether it
> builds against the latest MicroManager, that’d be helpful to spot
> problems early on. For testing we could later add tests.

Sure. On the other hand, what we *really* need to test against is the 
user installation as per the update site. And that requires regression 
tests, otherwise we think all is right when it might well be thoroughly 
broken.

> I haven’t used Jenkins too deeply so far, but the setup of a
> Maven build with Travis is quite easy.

I know that. The same is true for Maven projects in Jenkins: New job, 
type in the Git URL, build step "Maven", done.

So why do I say it is complicated?

Because we depend on stuff that is not available via Maven repositories. 
And worse, that stuff that is on Maven repositories is not necessarily 
what is present on the OpenSPIM update site. And all that needs to be 
cobbled together and... yep, you guessed it: regression-tested.

Ciao,
Johannes