On Jul 16, 2012, at 4:51 AM, Aristedes Maniatis wrote:

> Ignore the sqlite, that's never worked. I suspect it needs to be taken out 
> entirely and removed from any mention of a supported Cayenne DB, but I'm not 
> sure.

Well, it is supported as much as we can, just that it has a much smaller 
feature set. So IMO we shouldn't disavow it, but I am +1 for dropping it from 
Jenkins test matrix. 

> Also ignore the "true" column. That is for case-insensitive collation and 
> isn't yet set up properly. Dima, since you have recently been doing some work 
> in that area, perhaps you'd like to see what is required to get those tests 
> working?

Yeah, we either need to fix those or disable "true" from the test matrix 
temporarily.

> I also notice that all the Java 1.5 builds are failing. Has this support 
> already been removed or does this point to a problem?

Need to investigate. Could be one of those implicit ORDER BY assumptions or 
some other trivial thing in the tests... Or not..


> We have one failed Java 7 build. No idea why 
> test(org.apache.cayenne.merge.SetAllowNullToDbTest) is failing and only in 
> Derby under Java 7.
> 
> https://builds.apache.org/job/Cayenne-trunk/cayenneTestConnection=derby,collationCi=false,jdk=JDK%201.7%20%28latest%29,label=Ubuntu/740/console

Same thing... Could be ambiguity in the tests. Need to take a look.

> Finally, I created  a new set of builds for our 3.1 release branch.
> 
>  https://builds.apache.org/job/Cayenne-31/
> 
> We should ensure that Java 5 builds properly there at least.

Awesome! Will see what we need to fix.

Andrus

Reply via email to